个人工作站 › 调查报告 ← 返回报告清单
Pi Changelog2026-10-21 快照

pi 更新履历
0.10.0 → 1.4.0

十一个月、290 个版本、3173 条变更。从 2025-11-25 的「TUI + 七个工具」起步，到 2026-10 的 codemode、MCP、虚拟模型与缓存优化。先按 1.x 每个大版本给一张速览（中文一句 + 英文原文），再切成六段讲历史，然后把 121 条破坏性变更单列成表，最后附一份全量可检索的版本归档。

数据源：本机 node_modules/@earendil-works/pi-coding-agent/CHANGELOG.md（1.4.0，639 KB）
抓取方式：解析 markdown 后整页内联生成，条目原文为英文，分期说明与本机影响为中文整理。

总览

Overview

290 个版本按类别拆开的条目数分布，以及最直观的几个数：发版最密集的是 2025 年 12 月（一个版本号体系里能连发 75 版），最近一个月回落到每周 1–2 版。

版本总数 290 0.10.0 → 1.4.0
时间跨度 11 个月 2025-11 → 2026-10
条目总数 3173 按分类拆分
破坏性变更 121 分布于 50 个版本
最忙一个月 75 2025-12 发了 75 个版本
最近一版 1.4.0 2026-10-21

发布节奏与条目构成

Cadence

柱高＝当月版本数。0.x 阶段是把功能快速砸出来，1.0 之后节奏明显降下来，改动转向「省 token、少掉缓存、少出错」。

13 25/11
75 25/12
58 26/01
31 26/02
21 26/03
25 26/04
16 26/05
16 26/06
12 26/07
5 26/08
9 26/09
9 26/10

0.x 阶段几乎「一天一版」，2025-12 一个月发了 75 个版本；进入 1.x 后节奏落到每周 1–2 个功能版。

332新功能715新增228变更1601修复121破坏性变更13移除163其他

1.x 大版本速览

Major Releases

只收 1.x：5 个大版本各一张卡，每条新功能写成一句中文，下面紧跟 CHANGELOG 的英文原文（可直接搜关键词）。0.x 一天一版的细节往下看分期与全量归档。

1.4 2026-10-21 · 1.4.0 新功能 5 共 42 条

把两个「撞上就报错」的瓶颈改成排队 / 用满模型额度：codemode 满载不再失败，长思考也不再让摘要失败。

  • 分类器收图：models.classify() 在 typesafe / Cloudflare Workers AI / OpenRouter 上接受 imagesImage input for classifiers: codemode models.classify() accepts images for TypeSafe's Jev and Cloudflare's Clef models on typesafe, cloudflare-workers-ai, and OpenRouter. See Use classifier models and Codemode.
  • models.json 有了按 schema 生成的字段总参考：provider、model、覆盖项、兼容字段Models Configuration reference: every provider, model, model override, and compatibility field of models.json, generated from its schema. See Models Configuration.
  • 压缩与分支摘要改用模型自己的输出上限，长思考不再把摘要顶到 token capSummaries without a fixed output cap: compaction and branch summaries use the model's output limit, so long reasoning no longer makes them fail. See Compaction.
  • codemode 满载时脚本排队等空闲 worker，PI_CODEMODE_MAX_WORKERS 可抬上限Codemode waits for a free worker: parallel scripts and in-process subagents queue instead of failing, and PI_CODEMODE_MAX_WORKERS raises the limit. See Codemode limits.
  • 支持 Rex 终端：OSC 8 超链接、真彩色、内联图片Rex terminal support: OSC 8 hyperlinks, truecolor, and inline images.
1.3 2026-10-15 · 1.3.0 新功能 5 共 32 条

主题是「缓存不失效」落细：GPT-6 中途改档不掉缓存，换模型清掉旧思考，resume 保留会话中启用的工具。

  • /share 上传到哪儿由 sessionShareDestination 定（auto / gist / radius）Choose where /share uploads: sessionShareDestination selects auto, gist, or radius. See Sessions and context and Export or share a session.
  • GPT-6（API key）上中途改思考档不再让提示缓存失效Cache-preserving thinking level changes on OpenAI GPT-6: changing the thinking level mid-session with an API key keeps the prompt cache.
  • Kimi Code 改走 Chat Completions：中途加工具不再毁缓存；覆盖 baseUrl 必须带 /v1Kimi Code over Chat Completions: tools added mid-conversation no longer invalidate the prompt cache. models.json base URL overrides must now include /v1. See Providers.
  • 换模型时丢掉上一个模型的思考块，不再当正文塞给新模型Clean model switching: thinking from a previous model is dropped instead of being shown to the new model as text.
  • resume 恢复会话时保留会话中临时启用的工具Resumed sessions keep their active tools: --tools +name/-name applies on top of them. See Tools.
1.2 2026-10-12 · 1.2.0 新功能 5 共 43 条 破坏性 4

配置开始「可校验」：settings / models / keybindings / themes 发布 JSON Schema；OpenRouter 按 key 列出可用模型与价格。

  • OpenRouter 按 key 列出可用模型、上下文大小、最大输出与价格（含区域端点）Per-key OpenRouter model lists: with an OpenRouter key, Pi lists only the models that key may use, with the key's context size, max output, and prices, including regional endpoints set in models.json. See OpenRouter.
  • settings.json / models.json / keybindings / themes 加 $schema，编辑器里可校验可补全Published configuration schemas: add $schema to settings.json, models.json, keybindings, and themes for editor validation and completion. See Settings, Models, Keybindings, and Themes.
  • PI_CLIPBOARD_OSC52=1 让远程会话（跳板、mux）也能写进本地剪贴板Clipboard copies over remote sessions: PI_CLIPBOARD_OSC52=1 sends copied text to a remote client's clipboard when Pi cannot detect the remote session, such as a multiplexer attached from another machine. See Pi Process Configuration.
  • codemode 工具声明带出输入约束（整数、上下限、默认值），行内显示耗时与总时长Codemode improvements: tool declarations show input constraints such as integer, minimum/maximum, and default, and codemode rows show elapsed and total time. See Codemode.
  • 扩展 triggerTurn 起的运行也走 before_agent_start 那套准备before_agent_start for extension-triggered runs: runs started by pi.sendMessage(..., { triggerTurn: true }) go through the same setup as prompts. See Events and concurrency.
破坏性变更 · 4
  • schema 移到 https://pi.dev/schemas/*.schema.json，主题 schema 改名 theme.schema.jsonPublished generated configuration schemas for models, settings, keybindings, and themes at https://pi.dev/schemas/*.schema.json. The theme schema moved to https://pi.dev/schemas/theme.schema.json; themes loaded during startup now receive the same schema validation as later reloads. Update theme $schema references, define reusable colors under vars, and remove unsupported metadata (#9880, #10751 by @christianklotz)
  • before_agent_start 对 triggerTurn 运行也触发，事件按 messageType 变成联合类型Changed before_agent_start to also fire for runs started by pi.sendMessage(..., { triggerTurn: true }) while idle. BeforeAgentStartEvent is now a union on messageType: "user" events carry prompt and images, and "custom" events carry the starting message and no prompt. Messages returned by handlers are added to both kinds of run, and pending deliverAs: "nextTurn" messages are delivered with the custom-message run. Handlers that should act only on user prompts must check event.messageType === "user" (#10267, #5581, #10739 by @davidbrai)
  • 运行启动期间（含 before_agent_start 处理中）会话算忙：ctx.isIdle() 为假，消息进队列Changed the session to count as busy while a run starts, including during before_agent_start handlers: ctx.isIdle() is false, pi.sendMessage() queues into the starting run instead of starting a second run that failed with "Agent is already processing", pi.sendUserMessage() requires deliverAs, and an abort requested meanwhile aborts the run once it starts (#10739 by @davidbrai)
  • pi mcp list --json 对提供资源的 server 只报 resources: true，不再给资源数量Changed pi mcp list --json to report resources: true for servers that offer resources instead of resources and resourceTemplates counts; Pi no longer lists resources when connecting (#10526)
1.1 2026-10-07 · 1.1.0 新功能 5 共 38 条

把运行状态暴露给终端与面板，并让 --tools 支持在默认工具集上做增量增删。

  • OSC 7501 上报运行状态：工作中 / 等对话框 / 等登录 / 完成 / 失败Program status reporting: terminals and agent dashboards that support OSC 7501 see whether Pi is working, blocked on a dialog or login, done, or failed. See Program status.
  • Claude Haiku 5.5（anthropic/claude-haiku-5-5），思考可到 xhigh / maxClaude Haiku 5.5: anthropic/claude-haiku-5-5, with adaptive thinking up to xhigh/max effort.
  • --tools 支持 +name / -name：在默认工具集上增删，而不是整份替换Adjust default tools with +name/-name: --tools entries like pi -t +codemode,-write change the default selection instead of replacing it. See Tools.
  • GPT-6 Luna 可当分类器模型；codemode 的 classify() 也能收图GPT-6 Luna and image classification: OpenAI's GPT-6 Luna is available as a classifier model through the Decisions API, and codemode's models.classify() accepts images for classifiers that support them. See Use classifier models.
  • llama.cpp 0.6.0+ 上的 Julia-1 / Laya / Kev / lev / OpenJev 原生当分类器跑Native llama.cpp decision models: Julia-1, Laya, Kev, lev, and OpenJev served by llama.cpp 0.6.0 or later run natively as classifiers through /v1/systemone. See Classification.
1.0 2026-10-01 → 2026-10-05 · 5 个版本 新功能 17 共 83 条 破坏性 1

第一个 1.x：全屏 TUI 成为默认、codemode 提示词瘦身约 40%，并补上图像生成、Radius 登录与 MCP OAuth 加固。

  • 1.0.4--tools / --exclude-tools 支持 * 通配，新增 --no-mcp 一次关掉 MCP 工具Tool patterns and --no-mcp: --tools and --exclude-tools accept * patterns, for example --tools read,codemode,'mcp__radius__*' keeps only one MCP server's tools. --tools now keeps MCP tools unless an entry starts with mcp__, and --no-mcp turns off MCP for one run. See Tools and MCP tools.
  • 1.0.4codemode 里 tools.read() 读图片返回 image 块，image() 能再显示出来Codemode persists images: tools.read() on an image file now gives back an image block that image() can show. See Call tools.
  • 1.0.3azure provider 也能接 Foundry Chat Completions 部署（起于 azure/deepseek-v4-pro）Azure Foundry Chat Completions — The azure provider (renamed from azure-openai-responses) now also serves Foundry Chat Completions deployments, starting with azure/deepseek-v4-pro. See Azure OpenAI.
  • 1.0.3image() 同时把图写到临时文件，并在结果里给出路径供后续拿用Codemode images saved to files — image() also writes each image to a temp file and names the path in the result, so later turns can copy or move generated images. See Generate images.
  • 1.0.2samplingParamsByThinkingLevel 按思考档给 OpenAI 兼容 API 配 temperature / top_pSampling by thinking level — samplingParamsByThinkingLevel in models.json sets sampling parameters such as temperature and top_p for each thinking level on OpenAI-compatible APIs. See Configure sampling by thinking level.
  • 1.0.1Nix flake：nix run github:earendil-works/pi/stable 直接跑最新版Nix flake — nix run github:earendil-works/pi/stable runs the latest release, and nix profile add github:earendil-works/pi/stable installs it. See Install pi.
  • 1.0.1.pi/mcp.json 与 /mcp 能按项目开关或改用户级 MCP server 的暴露Project overrides for MCP servers — .pi/mcp.json and /mcp can enable, disable, or change the exposure of a user-level server for one project. See Configure servers.
  • 1.0.1oauth.clientRegistration: "cimd"：授权服务器按文档 URL 认 pi，不必动态注册MCP Client ID Metadata Documents — oauth.clientRegistration: "cimd" lets authorization servers allow pi by its document URL instead of dynamic registration. See Authenticate with OAuth.
  • 1.0.1pi.registerToolRenderer() 能给还没注册的工具（如 resume 后的 MCP 工具）画调用Tool renderers for any tool — pi.registerToolRenderer() draws calls to tools that are not registered yet, such as MCP tools in resumed sessions. See Tool rendering.
  • 1.0.1Cloudflare Clef / Clef Flash 分类器可在 codemode 与扩展里用Cloudflare Clef classifiers — @cf/cloudflare/clef and @cf/cloudflare/clef-flash are usable from codemode scripts and extensions. See Use classifier models.
  • 1.0.0TUI 默认全屏；tuiMode: "regular" 回到终端原生 scrollbackFullscreen by default — The TUI now runs fullscreen. Set tuiMode to "regular" to keep the terminal's normal scrollback. See Terminal and display.
  • 1.0.0codemode 提示词省约 40%，报错改成「教模型怎么恢复」Leaner codemode — About 40% fewer prompt tokens, and errors that tell the model how to recover. See Codemode.
  • 1.0.0codemode 用会话凭据调 models.generateImages() 生成图片Image generation in codemode — Scripts call models.generateImages() with the session's credentials. See Generate images and Use image models.
  • 1.0.0/login 里一步登录 Radius 并配好它的 MCP serverRadius in /login — Sign in with Radius and set up its MCP server in one step. See Radius.
  • 1.0.0浏览器在另一台机器时，用复制码登录 AnthropicAnthropic copy code login — Sign in when the browser runs on another machine. See Authenticate interactively.
  • 1.0.0MCP OAuth 加固：authServerMetadataUrl、RFC 9207 iss 校验、按 server 存凭据、step-up 登录保 scopesMCP OAuth hardening — oauth.authServerMetadataUrl, RFC 9207 iss checks, credentials per server, and step-up sign-in that keeps granted scopes. See Authenticate with OAuth.
  • 1.0.0quietStartup: "header" 只留版本与快捷键提示Header-only quiet startup — quietStartup: "header" keeps the version and key hints and hides the rest. See Terminal and display.
破坏性变更 · 1
  • 1.0.3azure-openai-responses 改名 azure：auth.json / models.json / settings.json 里的 key 都要改Renamed the Azure provider from azure-openai-responses to azure. Rename the provider key in auth.json (or run /login again), in models.json, and in settings.json (defaultProvider, enabledModels patterns, and modelThinkingLevels keys). Sessions that used the old provider fall back to another model when resumed, and their prompt cache is not reused. The AZURE_OPENAI_* environment variables are unchanged (#9714 by @jsanter27)

六段分期

Six Eras

按版本号与日期切成六段，每段给一句主线、几条关键事实和代表版本。想快速了解「pi 是怎么长成现在这样的」，看这一节就够；具体某条改动去最后一节搜。

01

从零到能用：TUI 骨架与工具集

0.10.0 – 0.30.2 · 2025-11-25 → 2025-12-26 · 88 个版本 · 259 条

2025-11-25 首发，随后一个月内几乎天天发版。交互式 TUI、七个内置工具、会话续接这些基本盘都在这一段落定；凭据、设置、会话 ID 也随之定型。

  • 首发即含 read/write/edit/bash/glob/grep/think 七个工具、/export HTML 导出、/model 运行时切模型，接 Anthropic / OpenAI / Google 三家。
  • hook 与 custom tool 体系在 12 月内长出雏形，随后被 0.35.0 合并为 extensions（见下一段）。
  • 0.28.0 把 API key 从 settings.json 搬进 auth.json；0.29.0 把 /clear 改名 /new；0.29.1 统一出 /settings 菜单并支持 SYSTEM.md 替换系统提示。
  • 0.31.0 会话条目改用 8 位短 ID，同时修掉 edit 在 Windows 上的 CRLF 匹配失败与 UTF-8 BOM 报错 —— 本机使用史里第一次出现 Windows 专属修复。
0.10.0 Initial public release 初版：TUI + 七个工具 + HTML 导出。
0.28.0 凭据搬家 apiKeys 从 settings.json 移除，改存 ~/.pi/agent/auth.json（破坏性）。
0.29.1 统一 /settings 思考级别、主题、队列模式收进一个菜单；自动加载 SYSTEM.md。
0.31.0 Windows 修修补补 CRLF 行尾与 BOM 导致 edit 失败的经典坑被修。
02

扩展体系成形：hooks → extensions

0.31.0 – 0.50.9 · 2026-01-02 → 2026-02-01 · 60 个版本 · 652 条

一月份的工作几乎都在给「第三方能插手 pi 的每个角落」铺路：消息队列拆成 steer / followUp、hooks 与 custom tools 合并成 extensions、浮层 UI 组件、RPC 协议。同时也是 provider 大扩张的开始。

  • 0.32.0 把 queueMessage() 拆成 steer / followUp 两套语义（6 条破坏性变更），消息插入时机从此可分辨。
  • 0.35.0 是扩展体系的分水岭：hooks + customTools 合并为单一 extensions，目录 hooks/、tools/ → extensions/，commands/ → prompts/，flag --hook/--tool → --extension。
  • 0.39.0 起 ctx.ui.custom({ overlay: true }) 能画浮动模态层，扩展的 UI 能力从纯文本行升级到组件树。
  • provider 侧：Hugging Face（0.50.2）、Kimi For Coding（0.50.3）、Vercel AI Gateway、OpenAI Codex 官方支持（0.47.0）密集接入；0.50.4 补上 SSH / mosh 下的 OSC 52 剪贴板。
0.32.0 steer / followUp 队列 API 一分为二，queueMode 改名 steeringMode（破坏性）。
0.35.0 extensions 单轨 hooks 与 custom tools 合流；目录与 CLI flag 全部改名（破坏性）。
0.39.0 overlay 组件 ctx.ui.custom({ overlay: true }) 支持浮动模态 UI。
0.50.4 OSC 52 /copy 在远程连接下也能写进本地剪贴板。
03

平台化：项目优先、内置工具可被接管

0.51.0 – 0.65.2 · 2026-02-01 → 2026-04-06 · 53 个版本 · 515 条

二到四月，pi 从「一个能跑 agent 的终端」变成「一个能被改造的平台」：资源优先级改成项目优先、内置工具的渲染与参数可被扩展接管、扩展工具默认并行执行。破坏性变更也在 0.62.0 达到历史峰值（单版 14 条）。

  • 0.51.0 一次补齐 Android / Termux、Alpine ARM64 musl、Nix / Guix 支持，并加入 bash spawn hook（执行前改写命令）。
  • 0.55.0 资源优先级改为 项目优先（cwd/.pi 先于 ~/.pi/agent），扩展名冲突不再连坐卸载整个扩展。
  • 0.58.0 Claude Opus/Sonnet 4.6 上到 1M 上下文；扩展工具调用默认并行执行，只保留 tool_call 预检用于拦截。
  • 0.62.0 内置 read/write/edit/bash/grep/find/ls 变成可扩展的 ToolDefinition，渲染器可整体替换；同时引入统一的 sourceInfo 来源溯源 —— 这一版 14 条破坏性变更，是全程最贵的一次升级。
  • 0.63.0 起 edit 支持一次改多处（multi-edit），0.64.0 让老会话里的 oldText/newText 自动折成 edits[]；0.65.0 给出 session runtime API 与 defineTool()。
0.51.0 跨平台爆发 Termux、linux-arm64-musl、Nix/Guix 同时到位。
0.55.0 项目优先 资源查找顺序倒转，项目级 .pi 覆盖用户级配置（破坏性）。
0.62.0 内置工具可扩展 七个内置工具全部 ToolDefinition 化 + sourceInfo 溯源（14 条破坏性）。
0.65.0 runtime / defineTool 会话运行时 API 与独立工具定义助手。
04

换东家与稳态：模型层统一、项目信任

0.66.0 – 0.80.10 · 2026-04-08 → 2026-07-16 · 61 个版本 · 878 条

四月到七月的主题是「收拾屋子」：仓库与 npm scope 迁到 earendil-works、`models.json` 支持注释、`ModelRuntime` 统一模型与认证、项目级资源要经过信任才加载。同期还有图像生成、Windows ARM64 二进制、`max` 思考档。

  • 0.66.0 的启动公告与 0.74.0 的「仓库链接与包名迁移」标志项目从 badlogic/pi-mono、@mariozechner/* 迁到 earendil-works/pi、@earendil-works/*；0.73.1 的 pi update --self 提前为此做了自更新兼容。
  • 0.73.1 models.json 支持注释与尾逗号（JSONC），自建 provider 终于能写说明。
  • 0.74.1 接入图像生成 API、Together AI、Windows ARM64 独立二进制。
  • 0.79.0 引入 项目信任：启动前先问是否加载项目级 settings / 扩展 / 技能；footer 同时开始显示最近一次 prompt cache 命中率。
  • 0.80.8 ModelRuntime 把模型目录、认证、刷新收成一层；/model 能后台刷新已配置 provider 的实时目录。
0.71.0 砍掉 Gemini CLI 移除 Gemini CLI 与 Antigravity 内置 provider（破坏性）。
0.73.1 JSONC models.json 允许注释与尾逗号。
0.74.0 迁到 earendil-works 仓库与包 scope 全面改名，旧链接重定向。
0.79.0 项目信任 未知项目不再自动加载本地资源；footer 显示 cache 命中率。
0.80.8 ModelRuntime 模型目录 / 认证 / 刷新统一到一层 SDK。
05

全屏 TUI 与上下文工程

0.81.0 – 0.87.1 · 2026-07-21 → 2026-09-22 · 16 个版本 · 536 条

七月末到九月，界面与上下文两头同时动刀：全屏 TUI 上线（自带 12 条破坏性变更）、Markdown 里能画 Mermaid 与 LaTeX、`AGENTS.override.md` 可按目录覆盖上下文；prompt cache 从「看不见」变成可预热、可按模型给预算。

  • 0.81.0 支持连本地 llama.cpp router，搜索并下载 Hugging Face 模型；扩展可注册完整 provider。
  • 0.84.0 全屏 TUI 模式上线（运行时可切换）、Mermaid / LaTeX 渲染、AGENTS.override.md 按目录覆盖上下文；同版 12 条破坏性变更，集中在扩展渲染接口。
  • 0.84.3 为 Windows 加可选的原生 PowerShell 工具，并把自更新改成「先暂存校验、再原子切换」。
  • 0.86.0 prompt cache 可预热（长工具调用期间保活），/bug 支持带脱敏诊断上报，压缩预算可按模型配置。
  • 0.87.0 规范化会话上下文与扩展边界：用 context_with_system 改系统消息、思考块不再被当作普通文本传递。
0.81.0 本地模型 llama.cpp router 接入 + 完整 provider 扩展。
0.84.0 全屏 TUI 全屏模式 + Mermaid/LaTeX + 目录级上下文覆盖（12 条破坏性）。
0.84.3 Windows PowerShell 可选原生 PowerShell 工具，Windows 下执行更自然。
0.86.0 缓存预热 prompt cache warming、/bug、按模型配压缩预算。
0.87.0 上下文边界 会话上下文规范化 + context_with_system。
06

codemode 纪元与 1.x 稳定化

0.99.0 – 1.4.0 · 2026-09-29 → 2026-10-21 · 12 个版本 · 333 条

0.99.0 是整个版本史里最大的一次结构变化：codemode（模型写 JS 调工具）、MCP、虚拟模型、分类器模型、system 主题一起来了。此后 1.0 把全屏设为默认，1.0–1.4 的补丁几乎都在做同一件事：**更省 token、更少缓存失效**，再把 codemode 与压缩这两个瓶颈从「撞上就报错」改成「排队 / 用满模型额度」。

  • 0.99.0 codemode + MCP + virtual models + classifier models + system 主题；扩展事件里多出 provider_stream_event 这类可观测钩子。
  • 0.99.1 GPT-6.1 Sol 成为 OpenAI 默认模型；1.1.0 起 --tools 支持 +name/-name 增量调工具，并通过 OSC 7501 上报运行状态。
  • 1.0.0 全屏成为默认 TUI（tuiMode: "regular" 可回退）；codemode 提示词瘦身约 40%（GPT-5.6 单请求约 5,300 → 3,300 token），错误信息改成「教模型怎么恢复」。
  • 1.2.0 给 settings.json / models.json / keybindings / themes 发布 $schema，编辑器里可校验可补全；OpenRouter 按 key 列出可用模型与价格。
  • 1.3.0 把「缓存不失效」做到细处：GPT-6 上中途改思考档不掉缓存、换模型时丢掉上一个模型的思考块而非塞成文本、resume 保留会话中临时启用的工具。
  • 1.4.0 收尾两个硬失败：codemode 双 worker 满载时脚本改为排队（最多 60 秒，PI_CODEMODE_MAX_WORKERS 可抬上限），压缩 / 分支摘要改用模型自己的输出上限而不是固定额度（长思考不再撞「generation hit the token cap」）。
0.99.0 Codemode and MCP 模型写 JS 调工具、MCP 接入、虚拟/分类器模型、system 主题。
1.0.0 Fullscreen by default 全屏成为默认；codemode 提示词瘦身 40%。
1.1.0 --tools +name/-name 在默认工具集上做增量增删，不用整份重写。
1.2.0 $schema 配置文件发布 JSON Schema，编辑器校验与补全。
1.3.0 缓存友好 改思考档保缓存、换模型清思考块、resume 保留工具。
1.4.0 排队而非报错 codemode 满载排队 60s；摘要用模型输出上限；分类器收图。

破坏性变更清单

Breaking Changes

原文照录，按版本倒序。只看 1.x 的话，只剩 5 条（1.2.0、1.0.3）。

跨大版本升级时真正会咬人的是这 121 条。历史上最贵的三次是 0.62.0（14 条，内置工具 ToolDefinition 化）、0.84.0（12 条，全屏 TUI 与扩展渲染接口）、 0.32.0 / 0.35.0（队列 API 拆分、hooks→extensions 合并）。只关心从 0.x 上到 1.4.0 的话，按这张表倒序读一遍即可。

版本日期变更内容
1.2.0 2026-10-12 Published generated configuration schemas for models, settings, keybindings, and themes at https://pi.dev/schemas/*.schema.json. The theme schema moved to https://pi.dev/schemas/theme.schema.json; themes loaded during startup now receive the same schema validation as later reloads. Update theme $schema references, define reusable colors under vars, and remove unsupported metadata (#9880, #10751 by @christianklotz)
1.2.0 2026-10-12 Changed before_agent_start to also fire for runs started by pi.sendMessage(..., { triggerTurn: true }) while idle. BeforeAgentStartEvent is now a union on messageType: "user" events carry prompt and images, and "custom" events carry the starting message and no prompt. Messages returned by handlers are added to both kinds of run, and pending deliverAs: "nextTurn" messages are delivered with the custom-message run. Handlers that should act only on user prompts must check event.messageType === "user" (#10267, #5581, #10739 by @davidbrai)
1.2.0 2026-10-12 Changed the session to count as busy while a run starts, including during before_agent_start handlers: ctx.isIdle() is false, pi.sendMessage() queues into the starting run instead of starting a second run that failed with "Agent is already processing", pi.sendUserMessage() requires deliverAs, and an abort requested meanwhile aborts the run once it starts (#10739 by @davidbrai)
1.2.0 2026-10-12 Changed pi mcp list --json to report resources: true for servers that offer resources instead of resources and resourceTemplates counts; Pi no longer lists resources when connecting (#10526)
1.0.3 2026-10-05 Renamed the Azure provider from azure-openai-responses to azure. Rename the provider key in auth.json (or run /login again), in models.json, and in settings.json (defaultProvider, enabledModels patterns, and modelThinkingLevels keys). Sessions that used the old provider fall back to another model when resumed, and their prompt cache is not reused. The AZURE_OPENAI_* environment variables are unchanged (#9714 by @jsanter27)
0.87.0 2026-09-21 Removed the inherited shouldStopAfterTurn agent option. Use finishTurn and return { action: "end" } instead. finishTurn runs before turn_end but applies the decision afterward, and it also receives error and aborted responses; migrate normal-response predicates by returning undefined for those hard exits. See the @earendil-works/pi-agent-core changelog for a complete before-and-after example.
0.87.0 2026-09-21 Added ContextEditEntry to the exported SessionEntry union. TypeScript consumers with exhaustive entry switches must handle context_edit; use replacement: null for omission and a content replacement otherwise.
0.87.0 2026-09-21 Made SessionManager canonical for AgentSession provider context. Assigning session.agent.state.messages no longer replaces future request history; restore with SessionManager.inMemory(cwd, { id }, entries), navigate with session.navigateTree(), or append through session.sessionManager and call session.refreshContext().
0.87.0 2026-09-21 Expanded TurnEndEvent with required boundary fields and added AgentBeforeSettleEvent to the exported ExtensionEvent union. Consumers constructing events or exhaustively switching on ExtensionEvent must handle the new shapes. ExtensionRunner.emit() no longer accepts turn_end; host integrations dispatch actionable boundaries with emitBoundary(baseEvent, buildContext).
0.87.0 2026-09-21 Deferred runs requested from agent_settled handlers until all settled handlers finish. Handlers still observe ctx.isIdle() === true, but no longer see a reentrant agent_start during the same notification dispatch.
0.86.0 2026-09-19 Changed inherited pi-ai provider stream inputs from Context to normalized TranscriptContext values. Custom providers must read system prompts and tool declarations from context.messages with getCurrentSystemPrompt() and getCurrentTools(). See Custom Streaming API.
0.86.0 2026-09-19 Restricted inherited ToolCall.arguments and ToolResultMessage.details to JSON-compatible values, changed ToolResultMessage into a conditional type, and made JsonValue arrays readonly.
0.86.0 2026-09-19 user_bash now fails closed: errors or invalid defined results abort the command without invoking later handlers or executing locally. Return undefined to continue propagation; otherwise return { operations } or { result } (#9068).
0.84.3 2026-08-24 Renamed the inherited GoogleThinkingLevel type to GoogleApiThinkingLevel and added ResolvedGoogleThinkingLevel for normalized adapter levels.
0.84.0 2026-08-06 Renamed the inherited pi-ai ModelsStreamTransforms interface to ModelsRequestTransforms because its header transformation now applies to all authenticated provider requests.
0.84.0 2026-08-06 Changed JSON and RPC message_update events to emit only assistantMessageEvent deltas, removing the cumulative message and assistantMessageEvent.partial fields that caused quadratic output growth. Clients that need partial messages must assemble deltas between message_start and message_end; the latter remains authoritative (#7290).
0.84.0 2026-08-06 ModelRegistry.getApiKeyAndHeaders() now returns ProviderHeaders with string | null values and preserves null header-deletion markers. Extensions that inspect returned headers must handle null; extensions forwarding them to pi-ai streams should pass them through unchanged. This prevents placeholder OpenAI credentials from being sent through Cloudflare AI Gateway (#7030).
0.84.0 2026-08-06 Changed ModelRegistry.refresh() to accept ModelsRefreshOptions and return ModelsRefreshResult instead of discarding cancellation and provider errors.
0.84.0 2026-08-06 Changed ModelRuntime.setRuntimeApiKey() to accept auth cancellation options rather than catalog refresh options. Call refresh({ providers: [providerId], signal }) separately when remote freshness is required.
0.84.0 2026-08-06 Required config-form extension OAuth refreshToken(credentials, signal) callbacks to accept and honor a concrete abort signal.
0.84.0 2026-08-06 Replaced dynamic provider refresh context store access with the read-only context.stored snapshot and generation-checked context.publish() transaction.
0.84.0 2026-08-06 Replaced the inherited pi-agent-core harness session model with the v4 lane-based Session, SessionStorage, and SessionRepo APIs, including durable operation records, global facts, shared sequence numbers, and tree-scoped lane views.
0.84.0 2026-08-06 Promoted the inherited v2 session and AgentHarness API from pi-agent-core's experimental entrypoint to its default export and removed the experimental subpaths.
0.84.0 2026-08-06 Removed the inherited legacy JSONL and in-memory repository APIs. Use pi-agent-core's v4 JsonlSessionRepo or InMemorySessionRepo, both implementing the new SessionRepo contract.
0.84.0 2026-08-06 Added the inherited required pi-agent-core FileSystem.renameFile() operation for atomic JSONL publication; custom harness file-system implementations must provide same-filesystem replacement semantics (#7707 by @davidbrai).
0.84.0 2026-08-06 Replaced experimental remote-session list summaries with durable SessionMetadata; RemoteSession.sessions no longer exposes runtime phase, model, thinking, attachment, or lock state, which remains available from acquired SessionSnapshot values (#7708).
0.83.0 2026-07-29 Upgraded bundled TypeBox aliases to 1.3.7, removing deprecated APIs including Type.Base, Type.Awaited, Type.Promise, Type.AsyncIterator, Type.Iterator, Type.Options, and Value.Mutate, while fixing compiled validation of nullable array tool arguments. Extensions using removed APIs must migrate to supported TypeBox APIs. See Package Dependencies (#7243 by @petrroll).
0.80.8 2026-07-16 Replaced the SDK's CreateAgentSessionOptions.authStorage and modelRegistry options with the async modelRuntime option. AuthStorage and its storage backends are no longer exported; use ModelRuntime (or a custom pi-ai CredentialStore), or readStoredCredential() for one-off reads of auth.json.
0.80.8 2026-07-16 Removed redundant ModelRuntime.getAll(), find(), getSnapshot(), and getAuthOptions() projections. Use the pi-ai Models methods getModels(), getModel(), getProviders(), and checkAuth() directly.
0.80.8 2026-07-16 Replaced SDK request-auth assembly through ModelRegistry.getApiKeyAndHeaders() with ModelRuntime.getAuth(). Passing a provider ID returns provider-scoped auth; passing a model also resolves built-in, models.json, and extension model headers.
0.80.8 2026-07-16 Changed extension-facing ModelRegistry.refresh() from synchronous void to Promise<void> because models.json loading is asynchronous. Extensions must await it before making synchronous registry reads.
0.80.8 2026-07-16 Moved canonical dynamic catalog refresh to async ModelRuntime.refresh()/pi-ai Models.refresh(). Legacy extension OAuth modifyModels remains supported as a synchronous compatibility projection after credential initialization.
0.80.7 2026-07-14 Removed the openai-responses compat.sendSessionIdHeader flag from models.json. Session-affinity behavior is now controlled by compat.sessionAffinityFormat ("openai", "openai-nosession", or "openrouter"). Replace sendSessionIdHeader: false with sessionAffinityFormat: "openai-nosession" (#6496 by @petrroll).
0.75.0 2026-05-17 Raised the minimum supported Node.js version to 22.19.0.
0.73.0 2026-05-04 Switched the built-in xiaomi provider from Token Plan AMS to Xiaomi's API billing endpoint, and renamed its /login display from "Xiaomi MiMo Token Plan" to "Xiaomi MiMo". XIAOMI_API_KEY now refers to the API billing key from platform.xiaomimimo.com. Users on Token Plan should switch to the appropriate xiaomi-token-plan-* provider and set the corresponding env var (#4112 by @Phoen1xCode).
0.72.0 2026-05-01 Replaced compat.reasoningEffortMap in models.json and pi.registerProvider() model definitions with model-level thinkingLevelMap (#3208). Migration: move old mappings from compat.reasoningEffortMap to thinkingLevelMap. Use string values for provider-specific thinking values and null for unsupported pi levels that should be hidden and skipped by cycling. See docs/models.md#thinking-level-map and docs/custom-provider.md.
0.71.0 2026-04-30 Removed built-in Google Gemini CLI and Google Antigravity support. Existing configurations using those providers must switch to another supported provider.
0.70.0 2026-04-23 Disabled OSC 9;4 terminal progress indicators by default. Set terminal.showTerminalProgress to true in /settings to re-enable (#3588)
0.69.0 2026-04-22 Migrated first-party coding-agent code, SDK/examples/docs, and package metadata from @sinclair/typebox 0.34.x to typebox 1.x. New extensions, SDK integrations, and pi packages should depend on and import from typebox. Legacy extension loading still aliases the root @sinclair/typebox package, but @sinclair/typebox/compiler is no longer shimmed. This migration also picks up the new @mariozechner/pi-ai TypeBox-native validator path, so tool argument validation now works in eval-restricted runtimes such as Cloudflare Workers instead of being skipped (#3112)
0.69.0 2026-04-22 Session-replacement commands now invalidate captured pre-replacement session-bound extension objects after ctx.newSession(), ctx.fork(), and ctx.switchSession(). Old pi and command ctx references now throw instead of silently targeting the replaced session. Migration: if code needs to keep working in the replacement session after one of those calls, pass withSession to that same method and do the post-switch work there. In practice, move post-switch pi.sendUserMessage(), pi.sendMessage(), and command-ctx/session-manager access into withSession, and use only the ReplacedSessionContext passed to that callback for session-bound operations. Footguns: withSession runs after the old extension instance has already received session_shutdown, old cleanup may already have invalidated captured state, captured old pi / old command ctx are stale, and previously extracted raw objects such as const sm = ctx.sessionManager remain the caller's responsibility and must not be reused after the switch.
0.68.0 2026-04-20 Changed SDK and CLI tool selection from cwd-bound built-in tool instances to tool-name allowlists. createAgentSession({ tools }) now expects string[] names such as "read" and "bash" instead of Tool[], --tools now allowlists built-in, extension, and custom tools by name, and --no-tools now disables all tools by default rather than only built-ins. Migrate SDK code from tools: [readTool, bashTool] to tools: ["read", "bash"] (#2835, #3452)
0.68.0 2026-04-20 Removed prebuilt cwd-bound tool and tool-definition exports from @mariozechner/pi-coding-agent, including readTool, bashTool, editTool, writeTool, grepTool, findTool, lsTool, readOnlyTools, codingTools, and the corresponding *ToolDefinition values. Use the explicit factory exports instead, for example createReadTool(cwd), createBashTool(cwd), createCodingTools(cwd), and createReadToolDefinition(cwd) (#3452)
0.68.0 2026-04-20 Removed ambient process.cwd() / default agent-dir fallback behavior from public resource helpers. DefaultResourceLoader, loadProjectContextFiles(), and loadSkills() now require explicit cwd/agent-dir style inputs, and exported system-prompt option types now require an explicit cwd. Pass the session or project cwd explicitly instead of relying on process-global defaults (#3452)
0.65.0 2026-04-03 Removed extension post-transition events session_switch and session_fork. Use session_start with event.reason ("startup" | "reload" | "new" | "resume" | "fork"). For "new", "resume", and "fork", session_start includes previousSessionFile.
0.65.0 2026-04-03 Removed session-replacement methods from AgentSession. Use AgentSessionRuntime for newSession(), switchSession(), fork(), and importFromJsonl(). Cross-cwd session replacement rebuilds all cwd-bound runtime state and replaces the live AgentSession instance.
0.65.0 2026-04-03 Removed session_directory from extension and settings APIs.
0.65.0 2026-04-03 Unknown single-dash CLI flags (e.g. -s) now produce an error instead of being silently ignored.
0.64.0 2026-03-29 ModelRegistry no longer has a public constructor. SDK callers and tests must use ModelRegistry.create(authStorage, modelsJsonPath?) for file-backed registries or ModelRegistry.inMemory(authStorage) for built-in-only registries. Direct new ModelRegistry(...) calls no longer compile.
0.63.0 2026-03-27 ModelRegistry.getApiKey(model) has been replaced by getApiKeyAndHeaders(model) because models.json auth and header values can now resolve dynamically on every request. Extensions and SDK integrations that previously fetched only an API key must now fetch request auth per call and forward both apiKey and headers. Use getApiKeyForProvider(provider) only when you explicitly want provider-level API key lookup without model headers or authHeader handling (#1835)
0.63.0 2026-03-27 Removed deprecated direct minimax and minimax-cn model IDs, keeping only MiniMax-M2.7 and MiniMax-M2.7-highspeed. Update pinned model IDs to one of those supported direct MiniMax models, or use another provider route that still exposes the older IDs (#2596 by @liyuan97)
0.62.0 2026-03-23 Changed ToolDefinition.renderCall and renderResult semantics. Fallback rendering now happens only when a renderer is not defined for that slot. If renderCall or renderResult is defined, it must return a Component.
0.62.0 2026-03-23 Changed slash command provenance to use sourceInfo consistently. RPC get_commands, RpcSlashCommand, and SDK SlashCommandInfo no longer expose location or path. Use sourceInfo instead (#1734)
0.62.0 2026-03-23 Removed legacy source fields from Skill and PromptTemplate. Use sourceInfo.source for provenance instead (#1734)
0.62.0 2026-03-23 Removed ResourceLoader.getPathMetadata(). Resource provenance is now attached directly to loaded resources via sourceInfo (#1734)
0.62.0 2026-03-23 Removed extensionPath from RegisteredCommand and RegisteredTool. Use sourceInfo.path for provenance instead (#1734)
0.62.0 2026-03-23 RPC get_commands: replace path and location with sourceInfo.path, sourceInfo.scope, and sourceInfo.source
0.62.0 2026-03-23 SlashCommandInfo: replace command.path and command.location with command.sourceInfo
0.62.0 2026-03-23 Skill and PromptTemplate: replace .source with .sourceInfo.source
0.62.0 2026-03-23 RegisteredCommand and RegisteredTool: replace .extensionPath with .sourceInfo.path
0.62.0 2026-03-23 Custom ResourceLoader implementations: remove getPathMetadata() and read provenance from loaded resources directly
0.62.0 2026-03-23 command.path -> command.sourceInfo.path
0.62.0 2026-03-23 command.location === "user" -> command.sourceInfo.scope === "user"
0.62.0 2026-03-23 skill.source -> skill.sourceInfo.source
0.62.0 2026-03-23 tool.extensionPath -> tool.sourceInfo.path
0.61.0 2026-03-20 Interactive keybinding ids are now namespaced, and keybindings.json now uses those same canonical namespaced ids. Older config files are migrated automatically on startup. Custom editors and extension UI components still receive an injected keybindings: KeybindingsManager. They do not call getKeybindings() or setKeybindings() themselves. Declaration merging applies to that injected type (#2391)
0.61.0 2026-03-20 Extension author migration: update keyHint(), keyText(), and injected keybindings.matches(...) calls from old built-in names like "expandTools", "selectConfirm", and "interrupt" to namespaced ids like "app.tools.expand", "tui.select.confirm", and "app.interrupt". See docs/keybindings.md for the full list. pi.registerShortcut("ctrl+shift+p", ...) is unchanged because extension shortcuts still use raw key combos, not keybinding ids.
0.60.0 2026-03-18 Changed package startup behavior so installed unpinned packages are no longer checked or updated during startup. Use pi update to apply npm/git package updates, while interactive mode now checks for available package updates in the background and notifies you when updates are available (#1963)
0.59.0 2026-03-17 Changed custom tool system prompt behavior so extension and SDK tools are included in the default Available tools section only when they provide promptSnippet. Omitting promptSnippet now leaves the tool out of that section instead of falling back to description (#2285)
0.57.0 2026-03-07 RPC mode now uses strict LF-delimited JSONL framing. Clients must split records on \n only instead of using generic line readers such as Node readline, which also split on Unicode separators inside JSON payloads (#1911)
0.56.0 2026-03-04 Changed scoped model thinking semantics. Scoped entries without an explicit :<thinking> suffix now inherit the current session thinking level when selected, instead of applying a startup-captured default.
0.56.0 2026-03-04 Moved Node OAuth runtime exports off the top-level @mariozechner/pi-ai entry. OAuth login and refresh must be imported from @mariozechner/pi-ai/oauth (#1814).
0.55.0 2026-02-24 Resource precedence for extensions, skills, prompts, themes, and slash-command name collisions is now project-first (cwd/.pi) before user-global (~/.pi/agent). If you relied on global resources overriding project resources with the same names, rename or reorder your resources.
0.55.0 2026-02-24 Extension registration conflicts no longer unload the entire later extension. All extensions stay loaded, and conflicting command/tool/flag names are resolved by first registration in load order.
0.53.0 2026-02-17 SettingsManager persistence semantics changed for SDK consumers. Setters now update in-memory state immediately and queue disk writes. Code that requires durable on-disk settings must call await settingsManager.flush().
0.53.0 2026-02-17 AuthStorage constructor is no longer public. Use static factories (AuthStorage.create(...), AuthStorage.fromStorage(...), AuthStorage.inMemory(...)). This breaks code that used new AuthStorage(...) directly.
0.52.10 2026-02-12 ContextUsage.tokens and ContextUsage.percent are now number | null. After compaction, context token count is unknown until the next LLM response, so these fields return null. Extensions that read ContextUsage must handle the null case. Removed usageTokens, trailingTokens, and lastUsageIndex fields from ContextUsage (implementation details that should not have been public) (#1382 by @ferologics)
0.52.10 2026-02-12 Git source parsing is now strict without git: prefix: only protocol URLs are treated as git (https://, http://, ssh://, git://). Shorthand sources like github.com/org/repo and git@github.com:org/repo now require the git: prefix. (#1426)
0.52.7 2026-02-06 Changed models.json provider models behavior from full replacement to merge-by-id with built-in models. Built-in models are now kept by default, and custom models upsert by id.
0.52.6 2026-02-05 Removed /exit command handling. Use /quit to exit (#1303)
0.51.3 2026-02-03 RPC get_commands response and SlashCommandSource type: renamed "template" to "prompt" for consistency with the rest of the codebase
0.51.0 2026-02-01 Extension tool signature change: ToolDefinition.execute now uses (toolCallId, params, signal, onUpdate, ctx) parameter order to match AgentTool.execute. Previously it was (toolCallId, params, onUpdate, ctx, signal). This makes wrapping built-in tools trivial since the first four parameters now align. Update your extensions by swapping the signal and onUpdate parameters:
0.50.0 2026-01-26 Header values in models.json now resolve environment variables (if a header value matches an env var name, the env var value is used). This may change behavior if a literal header value accidentally matches an env var name. (#909)
0.50.0 2026-01-26 External packages (npm/git) are now configured via packages array in settings.json instead of extensions. Existing npm:/git: entries in extensions are auto-migrated. (#645)
0.50.0 2026-01-26 Resource loading now uses ResourceLoader only and settings.json uses arrays for extensions, skills, prompts, and themes (#645)
0.50.0 2026-01-26 Removed discoverAuthStorage and discoverModels from the SDK. AuthStorage and ModelRegistry now default to ~/.pi/agent paths unless you pass an agentDir (#645)
0.47.0 2026-01-16 Extensions using Editor directly must now pass TUI as the first constructor argument: new Editor(tui, theme). The tui parameter is available in extension factory functions. (#732)
0.44.0 2026-01-12 pi.getAllTools() now returns ToolInfo[] (with name and description) instead of string[]. Extensions that only need names can use .map(t => t.name). (#648 by @carsonfarmer)
0.43.0 2026-01-11 Extension editor (ctx.ui.editor()) now uses Enter to submit and Shift+Enter for newlines, matching the main editor. Previously used Ctrl+Enter to submit. Extensions with hardcoded "ctrl+enter" hints need updating. (#642 by @mitsuhiko)
0.43.0 2026-01-11 Renamed /branch command to /fork (#641)
0.43.0 2026-01-11 SessionManager.list() and SessionManager.listAll() are now async, returning Promise<SessionInfo[]>. Callers must await them. (#620 by @tmustier)
0.39.0 2026-01-08 before_agent_start event now receives systemPrompt in the event object and returns systemPrompt (full replacement) instead of systemPromptAppend. Extensions that were appending must now use event.systemPrompt + extra pattern. (#575)
0.39.0 2026-01-08 discoverSkills() now returns { skills: Skill[], warnings: SkillWarning[] } instead of Skill[]. This allows callers to handle skill loading warnings. (#577 by @cv)
0.38.0 2026-01-08 ctx.ui.custom() factory signature changed from (tui, theme, done) to (tui, theme, keybindings, done) for keybinding access in custom components
0.38.0 2026-01-08 LoadedExtension type renamed to Extension
0.38.0 2026-01-08 LoadExtensionsResult.setUIContext() removed, replaced with runtime: ExtensionRuntime
0.38.0 2026-01-08 ExtensionRunner constructor now requires runtime: ExtensionRuntime as second parameter
0.38.0 2026-01-08 ExtensionRunner.initialize() signature changed from options object to positional params (actions, contextActions, commandContextActions?, uiContext?)
0.38.0 2026-01-08 ExtensionRunner.getHasUI() renamed to hasUI()
0.38.0 2026-01-08 OpenAI Codex model aliases removed (gpt-5, gpt-5-mini, gpt-5-nano, codex-mini-latest). Use canonical IDs: gpt-5.1, gpt-5.1-codex-mini, gpt-5.2, gpt-5.2-codex. (#536 by @ghoulr)
0.35.0 2026-01-05 Settings: hooks and customTools arrays replaced with single extensions array
0.35.0 2026-01-05 CLI: --hook and --tool flags replaced with --extension / -e
0.35.0 2026-01-05 Directories: hooks/, tools/ → extensions/; commands/ → prompts/
0.35.0 2026-01-05 Types: See type renames above
0.35.0 2026-01-05 SDK: See SDK migration above
0.33.0 2026-01-04 Key detection functions removed from @mariozechner/pi-tui: All isXxx() key detection functions (isEnter(), isEscape(), isCtrlC(), etc.) have been removed. Use matchesKey(data, keyId) instead (e.g., matchesKey(data, "enter"), matchesKey(data, "ctrl+c")). This affects hooks and custom tools that use ctx.ui.custom() with keyboard input handling. (#405)
0.32.0 2026-01-03 Queue API replaced with steer/followUp: The queueMessage() method has been split into two methods with different delivery semantics (#403):
0.32.0 2026-01-03 Settings renamed: queueMode setting renamed to steeringMode. Added new followUpMode setting. Old settings.json files are migrated automatically.
0.32.0 2026-01-03 AgentSession methods renamed:
0.32.0 2026-01-03 Hook API signature changed: pi.sendMessage() second parameter changed from triggerTurn?: boolean to options?: { triggerTurn?, deliverAs? }. Use deliverAs: "followUp" for follow-up delivery. Affects both hooks and internal sendHookMessage() method.
0.32.0 2026-01-03 RPC API changes:
0.32.0 2026-01-03 Settings UI: "Queue mode" setting split into "Steering mode" and "Follow-up mode"
0.30.0 2025-12-25 SessionManager API: The second parameter of create(), continueRecent(), and list() changed from agentDir to sessionDir. When provided, it specifies the session directory directly (no cwd encoding). When omitted, uses default (~/.pi/agent/sessions/<encoded-cwd>/). open() no longer takes agentDir. (#313)
0.29.0 2025-12-25 Renamed /clear to /new: The command to start a fresh session is now /new. Hook event reasons before_clear/clear are now before_new/new. Merry Christmas @mitsuhiko! (#305)
0.27.0 2025-12-22 Session hooks API redesign: Merged branch event into session event. BranchEvent, BranchEventResult types and pi.on("branch", ...) removed. Use pi.on("session", ...) with reason: "before_branch" | "branch" instead. AgentSession.branch() returns { cancelled } instead of { skipped }. AgentSession.reset() and switchSession() now return boolean (false if cancelled by hook). RPC commands reset, switch_session, and branch now include cancelled in response data. (#278)
0.24.0 2025-12-19 Custom tools now require index.ts entry point: Auto-discovered custom tools must be in a subdirectory with an index.ts file. The old pattern ~/.pi/agent/tools/mytool.ts must become ~/.pi/agent/tools/mytool/index.ts. This allows multi-file tools to import helper modules. Explicit paths via --tool or settings.json still work with any .ts file.
0.24.0 2025-12-19 Hook tool_result event restructured: The ToolResultEvent now exposes full tool result data instead of just text. (#233)
0.23.3 2025-12-17 Hooks: turn_end event's toolResults type changed from AppMessage[] to ToolResultMessage[]. If you have hooks that handle turn_end events and explicitly type the results, update your type annotations.
0.23.0 2025-12-17 Hooks: Replaced session_start and session_switch events with unified session event. Use event.reason ("start" | "switch" | "clear") to distinguish. Event now includes entries array for state reconstruction.
0.20.0 2025-12-13 Pi skills now use SKILL.md convention: Pi skills must now be named SKILL.md inside a directory, matching Codex CLI format. Previously any *.md file was treated as a skill. Migrate by renaming ~/.pi/agent/skills/foo.md to ~/.pi/agent/skills/foo/SKILL.md.
0.16.0 2025-12-09 New RPC protocol: The RPC mode (--mode rpc) has been completely redesigned with a new JSON protocol. The old protocol is no longer supported. See C1 for the new protocol documentation and C2 for a working example. Includes RpcClient TypeScript class for easy integration. (#91)
0.14.0 2025-12-08 Custom themes require new color tokens: Themes must now include thinkingXhigh and bashMode color tokens. The theme loader provides helpful error messages listing missing tokens. See built-in themes (dark.json, light.json) for reference values.

与本机相关

Local Impact

下面这些是从这份履历里挑出来、对本机这套配置（pi 1.4.0、Windows、自建 wb2api provider、8 个 npm 包 + 1 个本地包 + 若干本地扩展、codemode 开启）真正有影响的条目。

1.0.0 已生效

全屏 TUI 是默认了

本机 settings.json 没有设 tuiMode，所以现在跑的就是全屏模式。想换回终端原生 scrollback："tuiMode": "regular"，或临时 --tui-mode regular。

0.99.0 可考虑

主题仍钉在 dark，没吃到 system 自适应

0.99.0 起默认主题是 system（从终端自己的调色板取色，终端明暗切换时重建）。本机钉了 "theme": "dark" —— 想跟着终端走就把这行删掉或设为 "system"。

1.3.0 无影响

自建 provider 的 baseUrl 要带 /v1

1.3.0 把 kimi-coding 换成 Chat Completions，models.json 里覆盖它的 baseUrl 现在必须带 /v1。本机 wb2api 等自定义 provider 是 openai-completions 且 baseUrl 已含 /v1（如 http://127.0.0.1:7863/v1），不受这条影响；但以后接别的 OpenAI 兼容端点时记得同一条规矩。

1.2.0 / 1.4.0 可考虑

models.json 现在能挂 $schema

1.2.0 给 settings.json、models.json、keybindings、themes 发布了 JSON Schema。本机 models.json 有十来个自定义 provider / 模型条目，加一行 $schema 就能在编辑器里补全与报错。1.4.0 另给了从 schema 生成的 Models Configuration 参考文档，并把 provider 级 compat 也接到分类器模型上（imageInput / sendImageOptInHeader）—— 本机 wb2api 这种「代理别人的服务」的端点正好用得上。

0.60.0 / 1.4.0 注意

pi update 只管 pi 自己，包不会顺带升级

0.60.0 起启动不再自动更新未锁定的 npm / git 包；0.79.7 起 pi update 默认只更新 pi 自身，--all 才连 packages 一起，--extensions 只升包。本机装了 8 个 npm 包（pi-mcp-adapter、pi-web-access、@juicesharp/rpiv-ask-user-question、pi-background-tasks、@plannotator/pi-extension、pi-agent-browser-native、context-mode、pi-subagents）+ 1 个本地路径包（~/.pi/packages/pi-tui），要定期 pi update --extensions。1.4.0 修了 managed-install.json 以 UTF-8 BOM 开头时 pi update 直接报 “Managed install marker is missing or invalid”（Windows PowerShell 5.1 重写文件会塞 BOM）—— 本机的 pi 就是 npm 全局装法，走的正是这条路径。

0.62.0 → 1.4.0 注意

本地扩展踩在改动最密的接口上

herdr-agent-state.ts（~/.pi/agent/extensions）、~/.pi/packages/pi-tui/extensions 里的 open-tui/ 与 wb-gateway.ts / wb-panel.ts / wb-model-picker.ts / wb-stats.ts、以及 @juicesharp/rpiv-ask-user-question 的 ask-user-question.ts，都直接用着扩展上下文与 UI 能力：0.62.0 内置工具渲染接口重写、0.80.4 生命周期钩子（agent_settled、before_provider_request）、0.87.0 上下文边界、1.2.0 triggerTurn 走 before_agent_start、1.3.0 扩展 overlay 的 done() 只关自己的浮层。1.4.0 把并发 dialog 从「后开的顶掉先开的、先开的 promise 永不 settle」修成可叠加（关掉当前浮层会退回上一个），并放开了扩展 import @earendil-works/pi-ai/models 子路径（clampThinkingLevel 这类）。升级后若扩展表现异常，先查这几处。

0.99.0 → 1.4.0 无影响

codemode 越用越便宜

codemode 自 0.99.0 起内置。1.0.0 把提示词压掉约 40% 并让报错自带恢复提示；1.1.0 起可用 --tools +codemode,-write 在默认工具集上增量增删；1.0.4 起 --tools/--exclude-tools 支持 * 通配。1.4.0 起两个 worker 都忙时脚本排队最多 60 秒而不是立刻失败（PI_CODEMODE_MAX_WORKERS 可抬上限），models.classify() 也能收图片。

1.4.0 已生效

长思考不再让压缩失败

1.4.0 起 generateSummary() / generateSummaryWithUsage() 忽略 reserveTokens，摘要请求直接吃模型自己的输出上限，compaction.reserveTokens 退回只管「什么时候开始压缩」；此前思考吃满固定额度会报 “generation hit the token cap”，长会话 + defaultThinkingLevel: "high" 时最容易撞。

0.75.5 → 1.4.0 已生效

Windows 这几条直接受益

0.75.5 文件工具在 Windows 上改用异步文件系统操作；0.76.0 修 Windows / JetBrains 终端里的编辑行为；0.84.3 加入可选的原生 PowerShell 工具（不用再靠 bash 转发）。1.4.0 修 Windows shell 探测：Microsoft Store 的 shell 别名不再被拒、不在 PATH 里的 PowerShell 7 能被找到、Windows 不装在 C:\Windows 时旧版 WSL bash.exe 启动器也能认出来。

0.80.6 / 1.3.0 说明

默认 thinking high，上面的档位都在

0.80.6 增加 max 档、0.80.7 起 Fable 5 支持 xhigh/max；1.3.0 在 GPT-6（API key）上中途改档不再掉缓存 —— 本机 defaultThinkingLevel: "high" 属于中间档，往上还能试 xhigh/max。

全量归档

Full Archive

290 个版本、3173 条变更，按月分组。点版本号展开条目；搜索框匹配版本号、日期与条目正文，按 / 可直接聚焦。

2026-109 个版本
1.4.0 2026-10-21 新功能 5新增 5变更 7修复 25 Image input for classifiers

新功能5

  • Image input for classifiers: codemode models.classify() accepts images for TypeSafe's Jev and Cloudflare's Clef models on typesafe, cloudflare-workers-ai, and OpenRouter. See Use classifier models and Codemode.
  • Models Configuration reference: every provider, model, model override, and compatibility field of models.json, generated from its schema. See Models Configuration.
  • Summaries without a fixed output cap: compaction and branch summaries use the model's output limit, so long reasoning no longer makes them fail. See Compaction.
  • Codemode waits for a free worker: parallel scripts and in-process subagents queue instead of failing, and PI_CODEMODE_MAX_WORKERS raises the limit. See Codemode limits.
  • Rex terminal support: OSC 8 hyperlinks, truecolor, and inline images.

新增5

  • Added a Models Configuration reference generated from the models.json schema, listing all provider, model, model override, and compatibility fields (#10790)
  • Codemode models.classify() now accepts images for TypeSafe's Jev and Cloudflare's Clef models on typesafe, cloudflare-workers-ai, and OpenRouter, and lists Clef Omni and Jev 1.14 on Workers AI.
  • Provider-level compat in models.json now also applies to classifier models, with the System One fields imageInput and sendImageOptInHeader for proxies that serve another service's classifiers. Extension classifier models accept compat too.
  • Added MiniMax-M3.1-Flash to the minimax and minimax-cn providers, with thinking levels low through max (#11121)
  • Added support for the Rex terminal: OSC 8 hyperlinks, truecolor, and inline images (#11131, #11150 by @petrroll, #11183 by @nihaals)

变更7

  • Provider attribution headers now come from pi-ai's client identity. Extensions' before_provider_headers handlers no longer see the OpenRouter, NVIDIA NIM, and OpenCode attribution headers; returning null for one still removes it. Cloudflare requests send pi's detailed User-Agent instead of pi-coding-agent, as do GitHub release checks for tool downloads, and OpenCode requests send x-opencode-client whenever attribution is enabled, not only with a session.
  • generateSummary() and generateSummaryWithUsage() now ignore their reserveTokens parameter. Summary requests use the model's output limit instead.
  • Changed the default GitHub Copilot model from gpt-5.4, which Copilot retired, to gpt-6.1-sol.
  • Changed the default NVIDIA model to nvidia/nemotron-3.5-super-vl-120b-a12b (#10910 by @cv)
  • Changed the default Xiaomi model to mimo-v2.6-pro for the xiaomi and Xiaomi Token Plan providers, since mimo-v2.5-pro left the catalog.
  • Grok 4.3 and Grok 4.5 on the xai provider now offer the xhigh thinking level.
  • Sign in with ChatGPT: ChatGPT Business workspace usage-limit errors are not retried and tell the user to ask their workspace admin to increase the limit.

修复25

  • Fixed compaction and branch summaries failing with "generation hit the token cap" when reasoning used most of the fixed summary output limit. Summaries now use the model's output limit, and compaction.reserveTokens only controls when compaction starts (#9512, #9075).
  • Fixed models.schema.json missing descriptions for provider and model fields, so editors showed no hints for them (#10790)
  • Fixed pi update failing with "Managed install marker is missing or invalid" when managed-install.json starts with a UTF-8 BOM, which Windows PowerShell 5.1 adds when it rewrites the file.
  • Fixed HTML export showing literal asterisks for Markdown bold and italic next to CJK punctuation (#3353, #10154)
  • Fixed the terminal staying in mouse-tracking, focus-reporting, Kitty keyboard, and alternate-screen modes when pi exits without its normal terminal teardown, which flooded the shell prompt with escape sequences (#11170)
  • Fixed concurrent extension select/confirm/input/editor dialogs and inline custom() components in interactive mode: a second dialog replaced the first and the first promise never settled; closing the visible dialog now shows the previous one again (#11166, #9909)
  • Fixed pasting in the shell hanging after copying text in pi and suspending it with Ctrl+Z on Linux: the background wl-copy/xclip/xsel process that keeps serving the clipboard was suspended along with pi (#11179)
  • Fixed codemode scripts failing immediately when both codemode workers are busy, for example with parallel codemode calls or in-process subagents. Scripts now wait up to 60 seconds for a free worker, and PI_CODEMODE_MAX_WORKERS raises the limit of two workers (#11184)
  • Fixed extensions failing to load when importing @earendil-works/pi-ai/models (e.g. for clampThinkingLevel) by exposing the subpath through the extension loader aliases and virtual modules (#11156)
  • Fixed mcp.json files that start with a UTF-8 BOM failing to parse, which skipped all their MCP servers and made pi mcp add/pi mcp remove fail (#11191)
  • Fixed /compact <instructions> ignoring the instructions when summarizing the start of a turn that compaction splits. When the split turn was the only new content since the last compaction, the instructions were ignored entirely (#11058)
  • Fixed resumed sessions and /tree navigation leaving extension tools inactive when the branch recorded its tools before they were registered, such as tools from extensions installed after the session started. Extension tools the branch never added or removed are now active unless they are registered with defaultActive: false (#11062, #11075)
  • Fixed sessions breaking after recovery from a response cut off at the output limit: the truncated tool call was omitted but its tool results were kept, so providers rejected every later request (#11227, #11233 by @petrroll)
  • Fixed the grep tool following RIPGREP_CONFIG_PATH, where options such as --fixed-strings silently changed regex searches into literal ones. The tool now runs rg --no-config (#11203, #11230 by @petrroll)
  • Fixed Windows shell detection rejecting runnable Microsoft Store shell aliases (#9504 by @petrroll)
  • Fixed Windows shell detection missing PowerShell outside PATH and the legacy WSL bash.exe launcher when Windows is not installed in C:\Windows. pi now checks %ProgramFiles%\PowerShell\7, %SystemRoot%\System32\WindowsPowerShell\v1.0, and %SystemRoot% (#9501 by @petrroll)
  • Fixed GitHub Copilot sessions with large request bodies, such as many images, failing with HTTP 413 failed to parse request instead of compacting and retrying (#11016 by @petrroll)
  • Fixed the openai-codex provider offering gpt-5.5 and gpt-5.3-codex-spark, which always fail for ChatGPT accounts (#11173 by @lucasmeijer)
  • Fixed whitespace-only tool output (for example a bare \r\n from bash) making OpenAI Completions, OpenAI Responses, and Google requests fail with a 400 (#8720)
  • Fixed automatic retry not triggering for more gateway upstream failures, stream disconnected before completion errors relayed by Codex CLI proxies, and transient Google Vertex OAuth token refresh failures, and retrying errors because of digits inside request ids or byte counts (#10920, #11057, #10996)
  • Fixed OpenAI Responses models losing the prompt cache on every deferred tool load after a tool was removed (#10940)
  • Fixed Cloudflare Workers AI classifier pricing: Clef Flash costs $0.038 and Jev $0.042 per million input tokens instead of Jev being listed as free.
  • Fixed an unclosed $$ display-math block swallowing or corrupting every Markdown table after it (#11159)
  • Fixed Markdown bold and italic not rendering next to CJK punctuation (#10730 by @theBucky)
  • Fixed @ file autocomplete returning no results for queries containing regex metacharacters such as [ or ( (#10972)
1.3.0 2026-10-15 新功能 5新增 2变更 2修复 23 Choose where /share uploads

新功能5

  • Choose where /share uploads: sessionShareDestination selects auto, gist, or radius. See Sessions and context and Export or share a session.
  • Cache-preserving thinking level changes on OpenAI GPT-6: changing the thinking level mid-session with an API key keeps the prompt cache.
  • Kimi Code over Chat Completions: tools added mid-conversation no longer invalidate the prompt cache. models.json base URL overrides must now include /v1. See Providers.
  • Clean model switching: thinking from a previous model is dropped instead of being shown to the new model as text.
  • Resumed sessions keep their active tools: --tools +name/-name applies on top of them. See Tools.

新增2

  • Changing the thinking level mid-session on OpenAI GPT-6 models (API key) now keeps the prompt cache (#10853)
  • Added the sessionShareDestination setting to choose automatic, GitHub gist, or Radius uploads for /share (#10870 by @vegarsti)

变更2

  • Changed the kimi-coding provider to use OpenAI Chat Completions at https://api.kimi.com/coding/v1, the protocol used by the official Kimi Code client. K3, K3-256K, and Kimi For Coding now accept tools added mid-conversation without invalidating the prompt cache. models.json overrides of the kimi-coding base URL must now include /v1 (#10720)
  • Changed model switching to drop thinking blocks from earlier assistant messages produced by a different model, instead of sending them as plain text. The new model previously saw the old model's reasoning as part of its replies and could start writing its own reasoning into the visible answer (#10995)

修复23

  • Pinned internal Pi dependencies to exact release versions to prevent fresh installs from mixing incompatible packages (#10865)
  • Fixed resumed sessions dropping tools activated during the session. Resume now restores the session's active tools, with --tools +name/-name entries applied to them, unless --tools with tool names, --no-tools, or --no-builtin-tools selects tools explicitly (#10785)
  • Fixed stopping a codemode script or reaching its deadline waiting for a non-interruptible built-in to return on Bun (#10748)
  • Fixed repeated codemode timeouts or aborts accumulating surviving workers; admission now limits live workers to two across sessions, counting cancelled workers until actual exit (#10748)
  • Fixed --model choosing a provider by catalog order when several providers offer the match; Pi now prefers the provider with credentials and otherwise reports the ambiguity. --models and enabledModels add a model for each provider that offers it and warn about the ambiguity (#10160)
  • Fixed read line-range labels for numeric-string or null arguments in the TUI, session tree, HTML exports, and examples (#9887)
  • Fixed the Nix flake build failing when a provider switched APIs: the model catalog pin is now refreshed from the newest published catalog when its model API groups differ, and releases pin the newest catalog and build with it before tagging (#10985)
  • Fixed the Nix package putting its Node.js first on PATH, which made bash tool shells use Pi's Node instead of the user's node/npm. Pi's Node is now only a fallback (#10519)
  • Fixed prompt templates dropping empty quoted arguments ("" or ''), which shifted later positional arguments and prevented positional defaults from applying (#10867)
  • Fixed the Nix package build failing in its install check because it still expected chord and esbuild, which Pi no longer depends on
  • Fixed tool-call argument parsing dropping fields when a string contains an invalid \u escape such as \usepackage (#10868)
  • Fixed Anthropic and Bedrock Claude requests failing with a 400 when an error tool result contains images; the images are now sent after the tool results of the turn, so the model still sees them (#10794)
  • Fixed OpenCode Go Gemini models failing with a 404 by moving OpenCode Zen and OpenCode Go to the opencode.ai/inference endpoints, and added Google Generative AI support to the OpenCode Go provider
  • Fixed OpenCode Go MiniMax M2.7 failing with "Model does not support this protocol" by using the Anthropic Messages API
  • Fixed automatic retry not triggering for gateway Unable to forward upstream stream errors (#10920)
  • Fixed automatic retry not triggering for The service is temporarily unavailable errors, which also made compaction fail on the first attempt (#10965)
  • Fixed OpenAI-compatible responses showing thinking after the answer when a proxy such as LiteLLM sends reasoning and content in the same stream chunk (#10926)
  • Fixed Gemini 4 Argon requests on Google Vertex failing with "Thinking budget is not supported", and its thinking level being ignored on the Gemini API (#10890)
  • Fixed Xiaomi MiMo models failing with a 400 at the minimal thinking level; they now expose only the off and high levels (#10818)
  • Fixed openai-codex/gpt-6.1-sol using a 272K context window instead of the 500K window Codex now provides (#10840)
  • Fixed tool calls whose final arguments are malformed JSON silently running with only the fields parsed before the error; they now fail validation and show the model the raw arguments it sent (#10897)
  • Fixed Claude Haiku 5.5 and Claude Opus 5 not supporting the off thinking level (#10753)
  • Fixed done() of an extension's ctx.ui.custom({ overlay: true }) closing the topmost overlay instead of its own, which left the finished overlay on screen and another extension's dialog unresolved (#10648, #10999 by @masonc15)
1.2.0 2026-10-12 新功能 5破坏性变更 4新增 5变更 4修复 23移除 2 Per-key OpenRouter model lists

新功能5

  • Per-key OpenRouter model lists: with an OpenRouter key, Pi lists only the models that key may use, with the key's context size, max output, and prices, including regional endpoints set in models.json. See OpenRouter.
  • Published configuration schemas: add $schema to settings.json, models.json, keybindings, and themes for editor validation and completion. See Settings, Models, Keybindings, and Themes.
  • Clipboard copies over remote sessions: PI_CLIPBOARD_OSC52=1 sends copied text to a remote client's clipboard when Pi cannot detect the remote session, such as a multiplexer attached from another machine. See Pi Process Configuration.
  • Codemode improvements: tool declarations show input constraints such as integer, minimum/maximum, and default, and codemode rows show elapsed and total time. See Codemode.
  • before_agent_start for extension-triggered runs: runs started by pi.sendMessage(..., { triggerTurn: true }) go through the same setup as prompts. See Events and concurrency.

破坏性变更4

  • Published generated configuration schemas for models, settings, keybindings, and themes at https://pi.dev/schemas/*.schema.json. The theme schema moved to https://pi.dev/schemas/theme.schema.json; themes loaded during startup now receive the same schema validation as later reloads. Update theme $schema references, define reusable colors under vars, and remove unsupported metadata (#9880, #10751 by @christianklotz)
  • Changed before_agent_start to also fire for runs started by pi.sendMessage(..., { triggerTurn: true }) while idle. BeforeAgentStartEvent is now a union on messageType: "user" events carry prompt and images, and "custom" events carry the starting message and no prompt. Messages returned by handlers are added to both kinds of run, and pending deliverAs: "nextTurn" messages are delivered with the custom-message run. Handlers that should act only on user prompts must check event.messageType === "user" (#10267, #5581, #10739 by @davidbrai)
  • Changed the session to count as busy while a run starts, including during before_agent_start handlers: ctx.isIdle() is false, pi.sendMessage() queues into the starting run instead of starting a second run that failed with "Agent is already processing", pi.sendUserMessage() requires deliverAs, and an abort requested meanwhile aborts the run once it starts (#10739 by @davidbrai)
  • Changed pi mcp list --json to report resources: true for servers that offer resources instead of resources and resourceTemplates counts; Pi no longer lists resources when connecting (#10526)

新增5

  • Added reachesTools to tool definitions, so tools that reach other tools (such as codemode and tool_search) declare which tools a call may reach; the MCP extension uses it to wait for the servers a call needs and to decide whether a replacement codemode or tool_search tool can reach MCP tools
  • Added aliases to tool namespaces; codemode's describeNamespace() and the namespace option of searchTools() match them, and MCP namespaces use the server name
  • Added per-key OpenRouter model lists: with an OpenRouter key, Pi lists only the chat models that key may use, with the key's context size, max output, and prices, including guardrails and regional endpoints set as the base URL in models.json. The list is saved and used offline (#10353, #10672 by @davidbrai)
  • Added PI_CLIPBOARD_OSC52=1 to send copied text to the terminal with OSC 52 after the local clipboard write, so copies reach a remote client's clipboard when Pi cannot detect the remote session, such as multiplexer attaches from another machine or remote IDE daemons (#10826, #10835, #10114)
  • Added elapsed time while a codemode script runs and its total time once finished to the codemode tool row, like bash rows (#10708)

变更4

  • Changed the showHardwareCursor setting to use only the terminal cursor instead of also drawing Pi's reverse-video cursor
  • Changed codemode tool declarations to show input constraints such as integer, minimum/maximum, length, pattern, and default as @ tags on property comments, so scripts see them without the tool repeating them in descriptions (#10707)
  • Changed codemode declarations of MCP tools to spell out the result type (content, structuredContent, isError) instead of referring to CallToolResult<T>, which describeTool(), searchTools(), and ALL_TOOLS showed without defining it; the codemode description no longer includes the shared MCP types
  • Changed serializeConversation() (compaction, branch summaries, bug reports) to omit thinking blocks; replayed thinking is often not in the live context, so including it could make a summarization request larger than the context it summarizes and fail

修复23

  • Fixed codemode ignoring a // @options: line that is not the first line of the script, which ran the script without the requested timeout_ms; the call now fails with an error before the script runs (#10631)
  • Fixed Mistral streams being aborted mid-generation (e.g. during long thinking) after the request timeout; the timeout now applies only to waiting for response headers (#10609)
  • Fixed a symlinked AGENTS.md in a git worktree nested inside its main repo causing no project context file to load (#10681)
  • Fixed the code mode tool description claiming every failed tool call rejects; a tool that declares an output schema resolves with its value on failure too (e.g. bash with a nonzero exit)
  • Fixed runs started by pi.sendMessage(..., { triggerTurn: true }) removing system prompt sections added in before_agent_start after their first tool call, which changed the prompt mid-run and broke the provider prompt cache twice per run (#10267, #10739 by @davidbrai)
  • Fixed codemode blocking the process for seconds to minutes after scripts with many text() calls; joining the output items was quadratic (#10678)
  • Fixed print mode (-p) exiting 0 without a warning when the final response was cut off at the model's output token limit; it now prints the partial text, writes a warning to stderr, and exits 1 (#10799)
  • Fixed pi mcp login --help and the CLI docs describing --timeout as only the wait for the browser; it limits the whole sign-in (#10636)
  • Fixed ~ not being expanded in !pattern, +path, and -path entries of the extensions, skills, prompts, and themes settings, so entries like !~/skills/pdf matched nothing (#10765)
  • Fixed MCP servers with resources registering no tools until every resource was listed; Pi no longer lists resources when connecting, so /mcp and pi mcp list show that a server offers resources instead of counting them (#10526)
  • Fixed MCP tool calls failing when the server returns structured content that is not an object (an array, null, or a primitive, as MCP 2026-07-28 allows); codemode declarations of MCP tools without an output schema now type structuredContent as unknown instead of an object (#10416)
  • Fixed reading or writing auth.json on Windows failing with EPERM when the previous lock folder was still being deleted; Pi now retries briefly before reporting the error (#10801)
  • Fixed OpenRouter router models such as openrouter/auto showing a cost of -$1,000,000 per 1M tokens; OpenRouter's negative placeholder prices are now 0 (#10672 by @davidbrai)
  • Fixed Anthropic requests failing when the provider rejects the strict tool set for grammar size, tool count, or union limits; Pi now retries once with preferred tools non-strict (#10409)
  • Fixed "The upstream provider returned an error" not being retried as a transient provider error (#10842)
  • Fixed focused fullscreen overlays blocking mouse-wheel scrolling of the transcript (#10769)
  • Fixed fragmented terminal color replies leaking hex values into the prompt and opening the external editor (related reports: #10256, #10657)
  • Fixed inline images not rendering in Herdr since 1.1.0 (#10774 by @ogulcancelik)
  • Fixed key bindings that name a shifted symbol, such as + or ctrl++, not matching under the Kitty keyboard protocol or modifyOtherKeys (#9880 by @christianklotz)
  • Fixed the compaction status indicator embedded in a custom editor using the default muted colors instead of the editor border or thinking-level color, as the working indicator does (#10809, #10810 by @Jay-Madden)
  • Fixed MCP tool calls that were cancelled or timed out while waiting for an OAuth refresh still being sent to the server once the refresh finished (#10598)
  • Fixed MCP OAuth sign-in accepting an authorization response with an iss parameter that could not be checked because no authorization server metadata was discovered; sign-in now fails with an issuer mismatch. Reported by @AizenvoltPrime
  • Fixed codemode failing with a broken bridge error when Pi runs under node --watch on Node 24 or 26 (#10725)

移除2

  • Removed the unreleased experimental server, client, and plugin stack (src/experimental, src/cli/experimental) and its source-only ./client and ./experimental/plugin exports; none of it was part of the published package. The @earendil-works/pi-client, @earendil-works/pi-protocol, and @earendil-works/pi-server packages it used are removed and no longer released
  • Removed the dependency on @earendil-works/chord
1.1.0 2026-10-07 新功能 5新增 9变更 2修复 22 Program status reporting

新功能5

  • Program status reporting: terminals and agent dashboards that support OSC 7501 see whether Pi is working, blocked on a dialog or login, done, or failed. See Program status.
  • Claude Haiku 5.5: anthropic/claude-haiku-5-5, with adaptive thinking up to xhigh/max effort.
  • Adjust default tools with +name/-name: --tools entries like pi -t +codemode,-write change the default selection instead of replacing it. See Tools.
  • GPT-6 Luna and image classification: OpenAI's GPT-6 Luna is available as a classifier model through the Decisions API, and codemode's models.classify() accepts images for classifiers that support them. See Use classifier models.
  • Native llama.cpp decision models: Julia-1, Laya, Kev, lev, and OpenJev served by llama.cpp 0.6.0 or later run natively as classifiers through /v1/systemone. See Classification.

新增9

  • Added +name and -name entries to --tools, which change the default tool selection instead of replacing it, for example pi -t +codemode
  • Added durationMs to the tool render context and to tool_execution_end extension events: the recorded execution time of a final tool result (#10549)
  • Added outputPad to the tool render context (#10557 by @rwachtler)
  • Added OpenAI's GPT-6 Luna as a classifier model through the Decisions API, available with OPENAI_API_KEY (see Use classifier models)
  • Added images to codemode's models.classify() context, so classifiers that accept images, such as GPT-6 Luna, can judge them
  • Added program status reporting with OSC 7501: terminals and agent dashboards that support it see whether Pi is working, blocked on a dialog or login, done, or failed. PI_PROGRAM_STATUS=1|0 overrides detection (see Terminal setup) (#10607)
  • Added aborted to agent_settled session, extension, and JSON events, so integrations can tell a cancelled run from a finished one (#10607)
  • Added Claude Haiku 5.5 (anthropic/claude-haiku-5-5), with adaptive thinking up to xhigh/max effort and prompt caching on Bedrock
  • Added native llama.cpp decision models: Julia-1, Laya, Kev, lev, and OpenJev served by llama.cpp 0.6.0 or later are listed only as classifiers through /v1/systemone instead of as chat models (see Classification) (#10382)

变更2

  • Changed outputPad to also apply to ! command output, tool output, and summary blocks (#9946, #10557 by @rwachtler)
  • Changed pi mcp login --timeout to limit the whole sign-in, including requests to the authorization server, instead of only the wait for the browser (#10565)

修复22

  • Fixed bash and PowerShell results losing Took after reloading a session, and the live Took including wall-clock steps; both now show the recorded execution time (#10549)
  • Fixed managed installs keeping every old release; pi update now keeps only the new release and the one it updated from (#10392, #10511 by @davidbrai)
  • Fixed standalone binaries loading .env, .env.local, and .env.development from the launch directory into Pi's environment (#10473)
  • Fixed !! command headers losing their dim color once output arrives (#10557 by @rwachtler)
  • Fixed the codemode description not marking searchTools(), describeTool(), and describeNamespace() as async, which led models to serialize the unawaited promise as {} (#10555)
  • Fixed codemode output items running together, so models could not tell where one text() or console.log() output ended and the next began. With several text items, each now starts with a ==> text N/M <== line, and console calls follow the other output in one <console_output> block with one line per call
  • Fixed /mcp waiting for all servers to connect before opening; the manager now updates live and remains usable while enabling, reconnecting, or disabling servers (#10562)
  • Fixed images being dropped as "could not be resized" when running under node --watch on Node 24.19+ and 26.x, where Node posts its own messages on the image resize worker channel (#10527)
  • Fixed clipboard paste doing nothing in Termux, and failed copies there omitting the Termux:API install hint (#10391)
  • Fixed ! and RPC bash output keeping fragments of color codes, such as a stray m, when a code was split across output chunks (#10504)
  • Fixed MCP OAuth sign-ins that could not be cancelled while waiting on the authorization server and kept running after the session ended. The sign-in screen now cancels with Esc at every step, session shutdown aborts a running sign-in, and each request to the authorization server times out after 15 seconds (#10565)
  • Fixed shutdown waiting up to 15 seconds to refresh an MCP OAuth token that was about to expire, only to close the server's session (#10565)
  • Fixed the fullscreen text selection surviving session switches and other transcript rebuilds, which highlighted unrelated text in the new transcript (#9311, #10567 by @christianklotz)
  • Fixed OpenAI models on Bedrock ignoring the thinking level and always running at Bedrock's default reasoning effort (#9331, #10142 by @jsanter27)
  • Fixed model headers in models.json not overriding the originator and User-Agent headers of Codex requests (#10429 by @lucasmeijer)
  • Fixed server_busy and servers are currently busy provider errors ending the turn instead of being retried (#10543)
  • Fixed Mistral responses that end with finish_reason: "error" not being retried (#10487)
  • Reduced context-limit request failures by estimating input at 3.5 characters per token instead of 4 when calculating output limits (#10497)
  • Fixed Radius models disabled by an organization owner still being listed
  • Fixed Anthropic browser login failing with "localhost refused to connect" when port 53692 is reserved or in use, for example by Hyper-V/WSL port exclusions on Windows: login now falls back to a free loopback port (#10571)
  • Fixed session costs undercounting long prompts on models with prompt-length pricing tiers, such as Claude Haiku 5.5, Gemini 3.1 Pro, and GPT-5.4, through OpenCode, OpenCode Go, OpenRouter, Vercel AI Gateway, Google, MiniMax, and other providers
  • Fixed Markdown links not being clickable in Herdr (#10573)
1.0.4 2026-10-05 新功能 2新增 2修复 8 Tool patterns and --no-mcp

新功能2

  • Tool patterns and --no-mcp: --tools and --exclude-tools accept * patterns, for example --tools read,codemode,'mcp__radius__*' keeps only one MCP server's tools. --tools now keeps MCP tools unless an entry starts with mcp__, and --no-mcp turns off MCP for one run. See Tools and MCP tools.
  • Codemode persists images: tools.read() on an image file now gives back an image block that image() can show. See Call tools.

新增2

  • Added * patterns to --tools and --exclude-tools, for example --tools read,codemode,'mcp__radius__*'
  • Added --no-mcp to disable the built-in MCP support for one run

修复8

  • Fixed syntax highlighting losing colors after the first line of multiline strings and comments in fenced code blocks (#10143)
  • Fixed codemode scripts not receiving images from read: tools.read() now resolves to an image block for image files, which image() shows (#10251)
  • Fixed MCP OAuth sign-in failing with invalid_redirect_uri on servers with OpenID Connect client registration, such as mcp.modem.dev: pi now registers as a native client (#10493)
  • Fixed --tools removing MCP tools, which left pi --tools codemode without any MCP servers. --tools now keeps MCP tools unless an entry starts with mcp__
  • Fixed MCP session shutdown returning while a server was still connecting, leaving its transport open until the server answered or timed out (#10249)
  • Fixed system prompt rules and the skills hint naming tools hidden by prepareLoadout. Hidden tools are left out of the rules, the skills hint names no tool when the file reader is hidden, and codemode shows each tool's prompt guidelines with its declaration; ToolLoadout gains getPromptGuidelines() (#10343)
  • Fixed Bedrock requests that fail with The pending stream has been canceled after a stalled HTTP/2 connection not being retried automatically (#10379)
  • Fixed codemode scripts that patch built-ins (for example Array.prototype.toJSON = ...) crashing pi and leaving the tool call unsettled. Built-ins are now frozen before the script runs, so such patches have no effect (#10444)
1.0.3 2026-10-05 新功能 2破坏性变更 1新增 1变更 3修复 3 Azure Foundry Chat Completions

新功能2

  • Azure Foundry Chat Completions — The azure provider (renamed from azure-openai-responses) now also serves Foundry Chat Completions deployments, starting with azure/deepseek-v4-pro. See Azure OpenAI.
  • Codemode images saved to files — image() also writes each image to a temp file and names the path in the result, so later turns can copy or move generated images. See Generate images.

破坏性变更1

  • Renamed the Azure provider from azure-openai-responses to azure. Rename the provider key in auth.json (or run /login again), in models.json, and in settings.json (defaultProvider, enabledModels patterns, and modelThinkingLevels keys). Sessions that used the old provider fall back to another model when resumed, and their prompt cache is not reused. The AZURE_OPENAI_* environment variables are unchanged (#9714 by @jsanter27)

新增1

  • Added Azure Foundry Chat Completions deployments, starting with azure/deepseek-v4-pro (#9645, #9714 by @jsanter27)

变更3

  • Codemode image() now also saves each image to a temp file and names the path in the result, so later turns can copy or move generated images (#10310)
  • Output files (full text of truncated tool output, binary MCP resources, codemode images) are now readable only by the user
  • Home/End now always move the editor cursor to the line start/end; fullscreen transcript top/bottom moved to Ctrl+Home/Ctrl+End, which no longer move the editor cursor (#10314)

修复3

  • Fixed subscription logins such as Sign in with ChatGPT failing with refresh_token_invalidated after a request was cancelled during an OAuth token refresh
  • Fixed codemode failing for the rest of a session after a pnpm global update removed the running install, and added a restart hint when errors occur after pi was updated or removed on disk (#10439)
  • Fixed interactive sessions reporting a read EIO or setRawMode EIO crash (and asking to run /bug) when the terminal went away, e.g. after closing the window or resuming a suspended pi in a closed terminal
1.0.2 2026-10-04 新功能 1新增 1 Sampling by thinking level

新功能1

  • Sampling by thinking level — samplingParamsByThinkingLevel in models.json sets sampling parameters such as temperature and top_p for each thinking level on OpenAI-compatible APIs. See Configure sampling by thinking level.

新增1

  • Added samplingParamsByThinkingLevel to models.json for per-thinking-level sampling parameter overrides on OpenAI-compatible APIs. See Configure sampling by thinking level (#9776 by @mrexodia)
1.0.1 2026-10-03 新功能 5新增 6变更 2修复 13移除 1 Nix flake

新功能5

  • Nix flake — nix run github:earendil-works/pi/stable runs the latest release, and nix profile add github:earendil-works/pi/stable installs it. See Install pi.
  • Project overrides for MCP servers — .pi/mcp.json and /mcp can enable, disable, or change the exposure of a user-level server for one project. See Configure servers.
  • MCP Client ID Metadata Documents — oauth.clientRegistration: "cimd" lets authorization servers allow pi by its document URL instead of dynamic registration. See Authenticate with OAuth.
  • Tool renderers for any tool — pi.registerToolRenderer() draws calls to tools that are not registered yet, such as MCP tools in resumed sessions. See Tool rendering.
  • Cloudflare Clef classifiers — @cf/cloudflare/clef and @cf/cloudflare/clef-flash are usable from codemode scripts and extensions. See Use classifier models.

新增6

  • Added a copy key (app.message.copy, default ctrl+x) to OAuth sign-in screens in /login, /mcp, and /mcp login, which copies the sign-in URL when the browser cannot be opened or the wrapped link cannot be selected.
  • Added oauth.clientRegistration: "cimd" for MCP servers, which identifies pi with its Client ID Metadata Document on pi.dev instead of dynamic client registration, so authorization servers can allow pi by URL (#10302)
  • Added project overrides for user-level MCP servers: a .pi/mcp.json entry without command or url sets only enabled, exposure, and toolExposure of the user-level server, and /mcp can enable or disable a server for the current project (#10277)
  • Added Cloudflare's Clef and Clef Flash classifier models to cloudflare-workers-ai, usable from codemode scripts and extensions (#10316 by @ndisidore, #10322 by @RealAlexandreAI)
  • Added pi.registerToolRenderer(), which chooses how calls to a tool are drawn, including tools that are not registered (#10285)
  • Added a Nix flake for macOS and Linux: nix run github:earendil-works/pi/stable runs the latest release, and nix profile add github:earendil-works/pi/stable installs it. See Install pi (#9137)

变更2

  • pi update on global npm installations now recommends migrating to the managed installation from the pi.dev installer, which pins all dependencies.
  • Anthropic tools added or redefined mid-conversation are now defined inline in the conversation, so redefining a tool under the same name keeps the prompt cache instead of resending the full tool list.

修复13

  • Fixed installations resolving vulnerable brace-expansion 5.0.9 by pinning brace-expansion 5.0.12 as a direct dependency (GHSA-q2hr-2g5m-vwhr, GHSA-qhr7-859c-m2p7, GHSA-6j4f-fj2g-mc7p) (#10288)
  • Fixed a trailing comma in --models adding an extra model to the model cycle (#10334)
  • Fixed a codemode script that prints in a loop crashing pi by running out of memory: a script fails once its output passes 16 Mi characters or 100000 items (#10283)
  • Fixed JPEG, GIF, and WebP images rendered by extensions through Image not appearing in Kitty, Ghostty, WezTerm, and Warp (#10292)
  • Fixed MCP tool calls in resumed sessions and HTML exports rendering fully expanded until their server connected, or for good if it never did (#10285)
  • Fixed fullscreen Kitty images collapsing to a one-row strip after scrolling in WezTerm (#10319)
  • Fixed "Selected model is at capacity" provider errors ending the turn instead of being retried (#10278)
  • Fixed Cloudflare AI Gateway Claude models failing with a 404 by using dashed model IDs (claude-opus-5-5 instead of claude-opus-5.5)
  • Fixed Sign in with ChatGPT continuing when its callback port is taken by another login, which made the browser show "OAuth state mismatch"; it now fails with a port-in-use error (#10265)
  • Fixed Amazon Bedrock OpenAI models costing requests above 272k input tokens at the short-context rate; Bedrock models now include the pricing tiers listed on models.dev (#10326)
  • Fixed Amazon Bedrock Claude requests failing with "Invalid signature in thinking block" after the system prompt or tools changed (#10324)
  • Fixed Together DeepSeek V4 Pro losing its thinking level controls after Together renamed it to deepseek-ai/DeepSeek-V4-Pro-0813 (#10336 by @cv)
  • Fixed the default NVIDIA model pointing at nvidia/nemotron-3-super-120b-a12b, which NVIDIA no longer serves; the default is now nvidia/nemotron-3-ultra-550b-a55b

移除1

  • Removed npm-shrinkwrap.json from the published package. npm installations no longer pin transitive dependencies, and library consumers can now override them. Use the pi.dev installer for pinned installations (#5653)
1.0.0 2026-10-01 新功能 7新增 4变更 8修复 13 Fullscreen by default

新功能7

  • Fullscreen by default — The TUI now runs fullscreen. Set tuiMode to "regular" to keep the terminal's normal scrollback. See Terminal and display.
  • Leaner codemode — About 40% fewer prompt tokens, and errors that tell the model how to recover. See Codemode.
  • Image generation in codemode — Scripts call models.generateImages() with the session's credentials. See Generate images and Use image models.
  • Radius in /login — Sign in with Radius and set up its MCP server in one step. See Radius.
  • Anthropic copy code login — Sign in when the browser runs on another machine. See Authenticate interactively.
  • MCP OAuth hardening — oauth.authServerMetadataUrl, RFC 9207 iss checks, credentials per server, and step-up sign-in that keeps granted scopes. See Authenticate with OAuth.
  • Header-only quiet startup — quietStartup: "header" keeps the version and key hints and hides the rest. See Terminal and display.

新增4

  • Added an oauth.authServerMetadataUrl setting for MCP servers that advertise a wrong OAuth authorization server or none. Pi uses the configured metadata document instead of discovery (#10172).
  • Added quietStartup: "header", which keeps the startup header with version and key hints but hides the model scope line and loaded-resource listing.
  • Added models.generateImages() to codemode scripts. It runs image models such as OpenRouter's with the session's credentials and returns base64 image blocks that image() attaches to the result; usage counts toward the session cost like models.classify(). Extensions can call ctx.modelRegistry.generateImages(). See Use image models.
  • Added a copy code login method to Anthropic /login for headless setups where the browser runs on another machine (#10194 by @lucasmeijer).

变更8

  • Changed the default TUI mode to fullscreen. Set tuiMode to "regular" or pass --tui-mode regular to keep the terminal's normal scrollback.
  • /login now offers "Sign in with Radius" at the top level, as the last option, with its status. After a Radius sign-in, /login offers to configure the Radius MCP server in the global mcp.json with "auth": { "provider": "radius" } and reloads. Cancelling a login returns to the menu it was started from.
  • The provider docs page is renamed to Providers, its "Cloud Providers" section is now "Provider Specific Config", and it documents Radius first.
  • MCP OAuth credentials are now stored per server name and URL, so MCP servers with the same URL can sign in with different accounts. Credentials stored by URL alone move to the first server that uses them (#10252).
  • Codemode costs far fewer prompt tokens: with the default tools and codemode active, a GPT-5.6 request shrinks from about 5,300 to 3,300 tokens. The codemode description lists the script globals in one line each and points to the new Codemode reference for the models API, which the model reads when it needs it. Declared tools say in one line how scripts call them and what the call resolves to, instead of repeating their full declaration, and the system prompt's codemode guidance and MCP server section are shorter.
  • Codemode errors now say how to recover: reading a tool or models member that does not exist names the close matches (tools.Bash suggests tools.bash), models.classify() and models.generateImages() reject malformed arguments with the expected shape, an unknown model points to models.getAvailableOfType(), an oversized store() value explains what the store is for, and a script that generates images without showing them gets a note. Scripts that probed for a tool with typeof tools.name must use "name" in tools.
  • /login and /logout now label providers without credentials as "not configured" instead of "unconfigured".
  • OAuth browser pages now show the color Pi logo.

修复13

  • Fixed MCP OAuth sign-in accepting an authorization response whose iss parameter names another authorization server; the code is now rejected before it is exchanged (RFC 9207).
  • Fixed MCP OAuth sign-in failing with Invalid scope when the token response contains "scope": "", and similar failures for other empty or null optional OAuth fields (#10266).
  • Fixed the sign-in URL printed by /mcp login not being clickable when it wraps (#10186).
  • Fixed --provider without --model being silently ignored and running the default model from another provider; it now fails with an error (#10236).
  • Fixed MCP servers that ask for more scope (insufficient_scope) requesting sign-in over and over. The new sign-in requested only the missing scopes, so the new token lost access the previous one had; it now keeps the granted scopes.
  • Fixed user messages in the transcript keeping two full-width copies of every rendered line; they keep one, with identical output.
  • Fixed /login and /logout labeling every OAuth sign-in, including Radius, as a subscription; only subscription-backed providers say "subscription", other OAuth sign-ins say "account".
  • Fixed the startup header logo rendering with gaps in Apple Terminal; it now shows a colored "Pi" with the version instead.
  • Fixed deferred MCP tools that tool_search loaded being dropped on resume and /reload even when their server reconnected before the next prompt, because the session restored its tools before the MCP servers reconnected.
  • Fixed the system theme making pastel palettes such as Catppuccin Frappe much more vivid; palette colors now keep their chroma (#10255, #10293 by @dgtlntv).
  • Fixed slash command autocompletion not triggering when the input starts with whitespace (#10218 by @haoqixu).
  • Fixed color bleeding past mouse selections and search highlights in fullscreen mode when a styled token ends at the highlight boundary (#10169).
  • Fixed memory retained per rendered message in the transcript; a long assistant message keeps about a fifth of the heap it kept before.
2026-099 个版本
0.99.2 2026-09-30 新功能 4新增 5变更 3修复 14 MCP servers stay out of the way: servers with the default co

新功能4

  • MCP servers stay out of the way: servers with the default codemode exposure are no longer listed in the codemode description and no longer block the first prompt. They appear in a short system prompt section, and scripts find their tools with searchTools() and describeNamespace(). See Control tool exposure.
  • More MCP authentication options: oauth.clientName for servers that only accept known OAuth clients, and "auth": { "provider": "<provider>" } to authenticate HTTP servers with a provider's /login token. See Authenticate with OAuth.
  • Anthropic workload identity federation from the Anthropic SDK environment variables. See Use an API key from the environment.
  • /reload enables tools newly added to the defaultTools setting. See Tools.

新增5

  • Added a description field for MCP servers (pi mcp add --description), shown with the server in the system prompt and used to rank its tools in tool search, and a describeNamespace(name) codemode helper that returns a namespace's instructions and tool names. describeNamespace() and searchTools() accept a namespace as mcp__dev-radius, mcp__dev_radius, dev-radius, or dev_radius.
  • Added an oauth.clientName setting for MCP servers (pi mcp add --oauth-client-name) to change the client name sent during OAuth client registration, for servers that only accept known clients (#10226).
  • Added "auth": { "provider": "<provider>" } for HTTP MCP servers to send a provider's current /login token as the bearer token instead of using MCP OAuth. The token is read on every request, so provider refreshes apply. Only allowed in the global mcp.json and from extensions, and requires https except on loopback hosts.
  • Added Anthropic workload identity federation from the ANTHROPIC_FEDERATION_RULE_ID, ANTHROPIC_ORGANIZATION_ID, and ANTHROPIC_IDENTITY_TOKEN_FILE environment variables (see Providers) (#10177, #10242 by @philfreo).
  • /reload now enables tools newly added to the defaultTools setting. Tools removed from it stay enabled, tools turned off during the session stay off unless newly added, and --tools, --no-tools, and --no-builtin-tools still override the setting (#10245).

变更3

  • MCP servers with the default codemode exposure no longer appear in the codemode description; scripts find them with searchTools(). codemode-deferred is now an alias for codemode. Use direct exposure for tools the model should see without searching (#10212).
  • The codemode description no longer includes deferred tools, tool counts, or MCP server instructions, so it no longer changes when MCP servers connect or change their tools. The tool_search description no longer lists the servers whose tools it can load, for the same reason. Servers are listed instead in an mcp_servers system prompt section with a one-line summary, updated at the start of each prompt; a changed section is appended to the conversation. Scripts read server instructions with describeNamespace() (#10212).
  • The first prompt no longer waits for MCP servers without direct tools. They connect in the background and are waited for when a codemode script names them, a script searches tools, or tool_search runs (#10212).

修复14

  • Fixed new sessions intermittently ignoring the saved default model, or warning that no models are available, when it belongs to an extension-registered native provider with a stored credential (#9962, #10190 by @davidbrai).
  • Fixed the /mcp sign-in URL not being clickable when it wraps across lines, by emitting it as a terminal hyperlink with a Cmd/Ctrl+click to open line like /login (#10186).
  • Fixed codemode image() accepting malformed base64 data or unsupported image types, which persisted an invalid image block that made every later provider request fail with HTTP 400 (#10215).
  • Fixed codemode failing to start its script worker from the standalone Windows executable (#10204).
  • Fixed prompt submission slowing down with session length, because resolving the session's model selection looked up the model catalog once per assistant message (#10198).
  • Fixed model lookups slowing down for providers with a refreshed pi.dev catalog, because merging remote catalog models took quadratic time.
  • Fixed the built-in-tool-renderer.ts and minimal-mode.ts extension examples removing the built-in tools' summaries and guidelines from the system prompt (#10072, #10193 by @christianklotz).
  • Fixed context overflow detection for Z.AI CN endpoint Prompt exceeds max length errors (#10208).
  • Fixed Anthropic requests failing when a tool schema uses keywords Anthropic strict tool use rejects, such as minimum/maximum; such tools are now sent non-strict (#9953).
  • Fixed provider retries firing immediately when a Retry-After header contains an unparseable date; they now use exponential backoff (#9571).
  • Fixed extension commands registered without a string name or handler crashing pi when typing /; the extension now fails to load with an error instead (#10054).
  • Fixed collapsed codemode and MCP tool results filling the screen when the output is one long line, such as minified JSON. Like bash output, the preview is now limited to wrapped lines instead of logical lines.
  • Fixed codemode.mode: "only" listing read, bash, edit, and write in the system prompt's tool list although requests only declare codemode (#10192).
  • Fixed codemode scripts calling the wrong MCP tool when two tool names differ only in - and _, such as read-file and read_file. Like in Codex, MCP tool and namespace names now replace - with _ (mcp__my-server__x is now mcp__my_server__x), colliding tools of a server all get a hash suffix, and server names that differ only in - and _ are rejected (#10239).
0.99.1 2026-09-29 新功能 1新增 1变更 1修复 1 GPT-6.1 Sol

新功能1

  • GPT-6.1 Sol — Available on OpenAI, Azure OpenAI, and OpenAI Codex, and now the default OpenAI Codex model. See Select a model.

新增1

  • Added GPT-6.1 Sol (gpt-6.1-sol) to the OpenAI, Azure OpenAI Responses, and OpenAI Codex providers.

变更1

  • Changed the default OpenAI Codex model to GPT-6.1 Sol (gpt-6.1-sol).

修复1

  • Fixed /login with OpenAI failing in the bundled release with a missing openai-chatgpt.js module error.
0.99.0 2026-09-29 新功能 5新增 20变更 11修复 29 Codemode and MCP

新功能5

  • Codemode and MCP — Connect MCP servers and let models run JavaScript that calls tools in parallel. See MCP Servers and Enable codemode.
  • System theme — Pi's colors now come from your terminal's own palette by default. See Use your terminal's colors.
  • Sign in with ChatGPT — Use a ChatGPT subscription with the OpenAI provider through /login openai. See Authenticate interactively.
  • Virtual models — Extensions can route each request to a different physical model. See Virtual Models.
  • Classifier models — Run Jev classifiers from codemode scripts, or use any llama.cpp model as a classifier. See How codemode works and Classification.

新增20

  • Added codemode, tool search, and MCP support as built-in extensions. The codemode tool runs model-written JavaScript in a QuickJS sandbox that calls pi's tools; enable it with defaultTools or --tools and configure it with codemode.mode and codemode.inlineBudget. tool_search finds tools that are not declared to the model and declares them. MCP servers over stdio or streamable HTTP, with OAuth, come from mcp.json (global, or per project once trusted) or pi.registerMcpServer() and are managed with /mcp and pi mcp add|remove|list|login|logout. See MCP Servers and Enable codemode (#10040).
  • Added extension tool APIs for orchestrating tools: exposure (direct, model-only, codemode, deferred, or hidden), namespace, annotations, outputSchema with structuredContent, isError results, prepareLoadout(), and ctx.executeTool() for nested tool calls, which emit events with parentToolCallId and are recorded as bounded nestedCalls on the calling tool's result. See Tool exposure.
  • Added a warning when an extension that registers the same tool, command, or flag replaces a built-in extension (#10174 by @cristinaponcela).
  • Added experimental virtual models: extensions register them with pi.registerVirtualModel() and pick a physical model and thinking level for each request. The footer shows the routed model, /session lists cost per physical model, and examples/extensions/jev-router.ts routes with the Jev classifier. See Virtual Models.
  • Added Sign in with ChatGPT to the OpenAI provider in /login, which uses a ChatGPT subscription with the OpenAI API. Pi stores a stable deviceId in the global settings for this login and omits it from bug reports.
  • Added the system theme, now the default, which derives pi's colors from the terminal's reported foreground, background, and ANSI palette and rebuilds them when the terminal switches between light and dark. See Use your terminal's colors.
  • Added #rgb, oklch(), and okhsl() colors and an optional appearance field to theme files, and theme.style(), theme.colors, and theme.appearance for extensions. See Themes and TUI.
  • Added a classifier model for every llama.cpp chat model, answered from next-token label probabilities. See Classification.
  • Added inherited Jev classifier models on OpenRouter, Cloudflare Workers AI, Vercel AI Gateway, and OpenCode Zen.
  • Added the fullscreenWheelScrollLines setting and /settings entry for fullscreen mouse-wheel scrolling. The default "auto" accelerates fast wheel spins outside local macOS terminals (#9758).
  • Added per-input disposition to successful RPC prompt, steer, and follow_up responses, AgentSession.steer()/followUp(), and RpcClient.prompt()/steer()/followUp(); RpcClient.prompt() also accepts streamingBehavior (#9098, #9803).
  • Added image generation to ModelRuntime: generateImages() with runtime-resolved auth (stored credentials, OAuth, runtime API keys, models.json headers), plus getModelsOfType(), getModelOfType(), getAvailableOfType(), getAllModels(), and getAllAvailable(). OpenRouter image models are listed under the openrouter provider and share its credential; an upstream ID can have separate chat and image entries. models.json providers and extension registrations without a model list keep built-in image generation. Extension model lists can include discriminated chat, image, and classifier entries with operation implementations; when supplied, they replace the provider catalog across every operation. Chat-facing reads (getModels(), getAvailableSnapshot(), the model picker) are unchanged.
  • Added classifier support to ModelRuntime, including classify(), classifier model accessors, runtime-resolved authentication, and the built-in TypeSafe jev-latest model.
  • Added types=chat,image,classifier to pi.dev model catalog requests so remote refreshes overlay every supported model type; entries of unknown model types are ignored.
  • Added the provider_stream_event extension event for observing parsed provider events before normalization, with an opt-in /debug-provider example viewer (#9784, #9901 by @davidbrai).
  • Added a show/hide toggle (H) in HTML exports for custom messages marked display: false. Messages remain hidden by default and can also be revealed from the sidebar (#8896, #10020 by @rwachtler).
  • Added inherited Claude Sonnet 5.5 support for Anthropic with adaptive thinking and a 1M context window.
  • Added a Built-in section in pi config to disable the built-in mcp, llama.cpp, codemode, and tool-search extensions globally or per project, stored as -builtin:<name> in the extensions setting. SDK inline extensions opt in with builtin: true.
  • Added +name and -name entries to the defaultTools setting to add or remove tools without repeating the defaults, for example "defaultTools": ["+codemode"]. Project entries of this form apply on top of the user setting. Documented how to enable codemode without MCP and how to use classifier models such as Jev from codemode scripts.
  • Added the token usage and cost of codemode models.classify() calls to the codemode tool result, so they count toward the session cost; the codemode result shows each call's cost.

变更11

  • Switched the build from the TypeScript native preview to TypeScript 7.0 with an ES2024 target, and replaced tsx with Node's built-in type stripping for running from source (#9965).
  • Removed the [Themes] section from the startup banner. Custom themes remain available in /settings, and theme conflicts are still reported.
  • Changed the startup header to show the pi logo with the version instead of the app name.
  • Changed the built-in dark and light themes to the revised pi colors, written in OKHSL.
  • Changed light/dark terminal detection to use the reported background color first, then the terminal's light/dark report, then COLORFGBG. The first-time setup no longer shows the detected appearance.
  • Renamed the inherited OpenAI Codex provider to "OpenAI Codex (legacy)"; Sign in with ChatGPT on the OpenAI provider supersedes it.
  • Changed inherited terminal detection to treat TERM=*-direct as truecolor.
  • Built-in extensions and tools are named builtin:<name> (for example builtin:mcp and builtin:read) in errors, diagnostics, RPC source info, and bug reports, instead of <inline:name> and <builtin:name>. Their slash commands no longer carry a [t] autocomplete tag.
  • --no-extensions also disables the built-in extensions, including the llama.cpp provider. Load one explicitly with -e builtin:<name>, for example pi -ne -e builtin:mcp.
  • Tool calls without a custom call renderer, including direct MCP tool calls, now show their arguments: as key=value pairs on the title line when collapsed and one key: value line per argument when expanded. MCP calls are titled server/tool and their results collapse to 5 lines.
  • bash and powershell structured results, which codemode scripts receive, now hold up to 1 MiB of output instead of the model-facing 2000 lines or 50KB, and add truncated and full_output_path. Longer output keeps its first and last 512 KiB. Empty output is "" instead of (no output).

修复29

  • Fixed X11 clipboard text being misidentified as an image when the clipboard owner accepts unadvertised image targets (#9786).
  • Prevented managed git packages from automatically installing Pi peer dependencies and added warnings for extension packages that list host-provided modules in dependencies (#9863).
  • Fixed pinned git extensions loaded with -e continuing to use the first downloaded commit after the ref changes (#9982).
  • Fixed RpcClient skipping the next event listener when a listener unsubscribes while handling an event, which could make waitForIdle() time out after collectEvents() (#9990).
  • Fixed full-file read calls rendering as :1 when models send null for omitted offset and limit (#9996).
  • Fixed new sessions being lost when pi exits before the first assistant response. The session file is now created when the first user message is sent (#10000).
  • Fixed unloaded llama.cpp autoload presets overwriting a cached runtime context window with the GGUF training context (#10077, #10158 by @cristinaponcela).
  • Fixed custom themes ignoring terminal.trueColor and other terminal capability overrides and rendering with 256 colors (#9973, #10039 by @christianklotz).
  • Fixed pasting files copied in Finder inserting the file icon image instead of the file paths; paths are quoted in bash mode (#9999, #10136 by @christianklotz).
  • Fixed the startup header, loaded resources, and chat notices keeping their old colors after a theme change.
  • Fixed the Fireworks default model pointing at the removed Kimi K2.6 model; it now defaults to Kimi K3.
  • Fixed the OpenCode Go default model pointing at the removed Kimi K2.6 model; it now defaults to Kimi K3.
  • Fixed the Together default model pointing at the removed Kimi K2.6 model; it now defaults to Kimi K3.
  • Reduced CPU use while streaming in long sessions and when previewing themes: the footer caches session usage totals, collapsed bash results cache their preview, and sanitizeBinaryOutput() no longer splits output into per-character arrays.
  • Fixed the usage of tools called through ctx.executeTool(), for example from codemode scripts, being dropped from the session cost; it is now added to the calling tool's result usage.
  • Fixed inherited /skill autocomplete appearing empty when loaded skill names did not contain the letters in skill (#9944).
  • Fixed inherited path and @ autocomplete not working after opening wrappers such as (, [, {, <, or a backtick.
  • Fixed inherited image stretching in terminals that use the Kitty graphics protocol (#8938, #9957 by @rwachtler).
  • Fixed inherited shell cursor staying hidden after exit when an extension closed an overlay during shutdown (#10026).
  • Fixed inherited keyboard input being lost after a mouse click in a /settings submenu closed it.
  • Fixed inherited 1-hour Anthropic cache writes through Vercel AI Gateway being priced at the 5-minute rate (#9210).
  • Fixed inherited model-level samplingParams being dropped by direct stream()/complete() calls on OpenAI-compatible APIs (#9506).
  • Fixed inherited Mistral GLM requests failing with "Expected at most one leading ThinkChunk" after empty content deltas (#9674).
  • Fixed inherited OpenAI Fast mode requests being priced at the standard rate (#10034).
  • Fixed inherited Mistral reasoning models ignoring the requested thinking level (#9678).
  • Fixed inherited OpenCode Zen and OpenCode Go qwen3.8-flash thinking being replayed as plain text on later turns (#10047).
  • Fixed inherited OpenAI Responses streams from servers that omit output_index, such as llama.cpp, running mixed-up tool calls; such streams now end with an error (#9974).
  • Fixed inherited Anthropic and OpenAI Codex browser sign-in waiting indefinitely after the provider redirected with an authorization error, and Anthropic sign-in failing when its callback port is in use.
  • Fixed inherited GitHub Copilot Claude Opus 5.5 offering unsupported thinking levels when upstream model metadata is incomplete.
0.87.1 2026-09-22 新功能 2新增 3变更 1修复 4 Latest frontier models

新功能2

  • Latest frontier models — Use Claude Opus 5.5, GPT-6 Sol, and GPT-6 Luna through supported providers, including GitHub Copilot. See Choose a Model.
  • Grok 4.7 by default for xAI — New xAI sessions now default to Grok 4.7. See Provider Authentication.

新增3

  • Added inherited Claude Opus 5.5, GPT-6 Sol, and GPT-6 Luna support for GitHub Copilot.
  • Added inherited GPT-6 Sol and GPT-6 Luna support for OpenAI API keys and OpenAI Codex subscriptions.
  • Added inherited Claude Opus 5.5 support for Anthropic with adaptive thinking and a 1M context window.

变更1

  • Changed the default xAI model to Grok 4.7.

修复4

  • Fixed split-turn compaction summaries being refused by Claude Fable 5.1 by clearly separating the conversation and using continuation-oriented instructions (#9908 by @davidbrai).
  • Fixed missing or invalid --mode values being silently ignored instead of reporting an error and exiting with a nonzero status (#9045).
  • Fixed inherited image-only user messages being rejected by some OpenAI-compatible providers because they included an empty text part (#9797).
  • Fixed inherited Anthropic OAuth requests reporting an outdated Claude Code version.
0.87.0 2026-09-21 新功能 3破坏性变更 5新增 5修复 11 Canonical session context and extension boundaries

新功能3

  • Canonical session context and extension boundaries — Edit model context without rewriting history and add actionable lifecycle hooks. See ContextEditEntry and extension events.
  • Full-transcript context extensions — Use context_with_system for per-request system-message transformations. See C1.
  • Per-model image input limits — Configure cache-safe image resizing per model for attachments, read, and tool-result images. See Image Input Limits.

破坏性变更5

  • Removed the inherited shouldStopAfterTurn agent option. Use finishTurn and return { action: "end" } instead. finishTurn runs before turn_end but applies the decision afterward, and it also receives error and aborted responses; migrate normal-response predicates by returning undefined for those hard exits. See the @earendil-works/pi-agent-core changelog for a complete before-and-after example.
  • Added ContextEditEntry to the exported SessionEntry union. TypeScript consumers with exhaustive entry switches must handle context_edit; use replacement: null for omission and a content replacement otherwise.
  • Made SessionManager canonical for AgentSession provider context. Assigning session.agent.state.messages no longer replaces future request history; restore with SessionManager.inMemory(cwd, { id }, entries), navigate with session.navigateTree(), or append through session.sessionManager and call session.refreshContext().
  • Expanded TurnEndEvent with required boundary fields and added AgentBeforeSettleEvent to the exported ExtensionEvent union. Consumers constructing events or exhaustively switching on ExtensionEvent must handle the new shapes. ExtensionRunner.emit() no longer accepts turn_end; host integrations dispatch actionable boundaries with emitBoundary(baseEvent, buildContext).
  • Deferred runs requested from agent_settled handlers until all settled handlers finish. Handlers still observe ctx.isIdle() === true, but no longer see a reentrant agent_start during the same notification dispatch.

新增5

  • Added append-only model-context edits. For example, sessionManager.appendContextEdit(entryId, null) omits one message from future provider context without changing raw history, usage, or UI history.
  • Added actionable turn_end and agent_before_settle extension boundaries. Return { entries: [...event.entries, draft], continue: true } to persist structural entries in order and ensure one next provider request without changing steering or follow-up scheduling.
  • Added retain-none compaction input: sessionManager.appendCompaction(summary, null, tokensBefore) stores the compaction's own ID as its kept boundary.
  • Added the context_with_system extension event, which runs after context handlers on the full transcript including system messages and sends its result verbatim. See C2.
  • Added per-model image resize profiles through inputLimits.images.resize in models.json, applied to file attachments, image reads, and tool-result images (#9631).

修复11

  • Fixed string context-edit replacements producing invalid assistant and tool-result message content instead of text blocks.
  • Fixed context-invisible boundary metadata and replacement edits causing newly appended or replaced input to be summarized before its first provider request.
  • Fixed edited-context accounting both discarding valid assistant usage captured after the latest context edit and reusing that usage after a later compaction made it stale.
  • Fixed selected error retries and final length/overflow recovery retaining abandoned model attempts in future provider context; post-run recovery omissions are now persisted without hiding raw transcript history or changing queue scheduling.
  • Fixed context handlers that filter or slice messages dropping the prompt and tool declarations, which after extension-driven compaction left requests without built-in tools or made Codex emit raw tool-call text. Handlers no longer see system messages; Pi restores the prompt and tool state after they run. See C1 (#9789, #9822).
  • Fixed /bug allowing uploads in offline mode while preserving local zip exports (#9841 by @christianklotz).
  • Fixed idle prompt-cache warming rebuilding expired caches when its timer or an extension decision is delayed.
  • Improved crash diagnostics with hints identifying loaded extensions that appear in the stack trace.
  • Fixed text files beginning with GIF being misclassified as images and omitted from read and CLI @file input (#9755).
  • Fixed malformed prompt template frontmatter being silently ignored instead of reported as a resource warning (#9830 by @christianklotz).
  • Fixed inherited unknown OpenAI-compatible Chat Completions endpoints receiving strict tool schemas unless they explicitly advertise support (#9816).
0.86.1 2026-09-20 新功能 1新增 1变更 1修复 5 Meta Muse provider

新功能1

  • Meta Muse provider — Sign in with Meta using /login meta or use META_API_KEY to access Muse Spark models. See Meta (Muse subscription).

新增1

  • Added Meta (Muse subscription) login via /login meta with automatic Model API key refresh, plus META_API_KEY support (#9096 by @xl0).

变更1

  • Enabled Node's persistent compile cache before loading the bundled CLI runtime, reducing repeat launch time.

修复5

  • Fixed /bug descriptions dropping line breaks from pasted diagnostics.
  • Fixed /bug hints appearing for user cancellations and retryable provider failures such as service unavailability.
  • Fixed clipboard copy failing in containers and WSL without WSLg by restoring the OSC 52 fallback when no display is available, and added a verified Windows clipboard backend for WSL (#9688).
  • Fixed inherited z.ai Prompt too long errors not being recognized as context overflow (#9805).
  • Fixed inherited Cerebras models advertising unsupported strict tool schemas, which caused HTTP 400 errors when strict and non-strict tools were mixed (#9804 by @EdenGottlieb).
0.86.0 2026-09-19 新功能 5破坏性变更 3新增 11变更 9修复 38移除 1 Prompt cache warming

新功能5

  • Prompt cache warming — Keep valuable prompt caches alive during long tool runs and optionally while idle using cost-aware refreshes. See Cache Warming.
  • Bug reporting — Report problems with /bug using redacted diagnostics, optional transcripts, or exported ZIP archives. See Reporting Bugs.
  • Transcript-aware prompt and tool updates — Preserve instruction and tool changes across resume and branch navigation while retaining cached prefixes. See C0.
  • Offline Radius model catalog — Select Radius models immediately, with cached and live catalogs overlaid when available. See Radius.
  • Per-model compaction budgets — Configure reserved and recent-token budgets by model. See Per-model overrides.

破坏性变更3

  • Changed inherited pi-ai provider stream inputs from Context to normalized TranscriptContext values. Custom providers must read system prompts and tool declarations from context.messages with getCurrentSystemPrompt() and getCurrentTools(). See Custom Streaming API.
  • Restricted inherited ToolCall.arguments and ToolResultMessage.details to JSON-compatible values, changed ToolResultMessage into a conditional type, and made JsonValue arrays readonly.
  • user_bash now fails closed: errors or invalid defined results abort the command without invoking later handlers or executing locally. Return undefined to continue propagation; otherwise return { operations } or { result } (#9068).

新增11

  • Added transcript-backed mid-conversation system prompt and tool changes so instruction and tool updates survive resume and branch navigation while preserving cached prefixes on supported models. See C0 and Entry Types (#9548).
  • Added inherited native deferred tool loading for Fireworks Messages models. Use ToolSearch or tool_search as the loader name for prompt-prefix deferral (#9323).
  • Added click toggling for branch summaries, compaction summaries, and skill invocation entries.
  • Added the public Radius model catalog for immediate and offline model selection, with cached and live gateway catalogs overlaid when available.
  • Added ctx.modelRegistry.stream() and streamSimple() for extension model calls through configured providers with resolved authentication (#8964).
  • Added per-model reserveTokens and keepRecentTokens settings through compaction.modelOverrides, with ordinary compaction settings as fallback (#8133).
  • Added compat.allowedFallbackModels configuration for overriding or disabling Anthropic server-side fallback models (#9294).
  • Added an unsubscribe function from pi.on() so extensions can drop event handlers. Handlers added or removed during a dispatch apply to later dispatches, not the current one (#8967).
  • Exported extension hook event and result types that were previously omitted from the package entry points (#9642).
  • Added /bug [description] to report a bug to the Pi developers. The report bundles environment, model, provider, extension, and settings metadata (secrets redacted), assistant message diagnostics from the session, optionally the session transcript, or a model-written summary of what went wrong instead. It is uploaded to Radius (no login required; attributed when logged in) or exported as a zip archive, and the report id is recorded in the session as a pi.bug-report entry. Crashes are recorded in ~/.pi/agent/crashes.json, announced once on the next start, and attached to the next report; unexplained errors and exhausted retries point at /bug once per session.
  • Added cost-aware prompt-cache warming during long tool runs and optionally while idle, with configurable modes, model cache-lifetime metadata, /session diagnostics, transcript notices, and the cache_warming_decision extension event. See Cache Warming (#9668).

变更9

  • Enabled Node's persistent compile cache before loading the bundled CLI runtime, reducing repeat launch time.
  • Made --resume session results appear progressively, using file modification times to prioritize all-folder loading and cancelling outstanding transcript reads after selection.
  • Reduced --continue startup time by checking candidate session headers in modification-time order and stopping after the newest matching session.
  • Replaced the external native clipboard dependency with bundled asynchronous macOS, Windows, and X11 helpers while preserving platform command and OSC 52 fallbacks (#9163).
  • Reduced inherited fuzzy search latency for long texts by using native substring search instead of scanning each character in JavaScript (#9267).
  • Moved compaction, branch summarization, and retry spinners into the editor border alongside the working indicator. Custom editors use the same embedding opt-in for all status spinners.
  • Enabled strict-prefer JSON-schema sampling by default for built-in read, bash, powershell, edit, and write tools, without requiring PI_EXPERIMENTAL. Extensions can re-register tool definitions with constrainedSampling: false.
  • Formatted Bash and PowerShell tool durations of at least one minute as minutes and seconds, with hours when needed (#9628).
  • Deferred the extension compiler and bundled virtual modules until a filesystem extension is loaded, reducing the baseline SDK import cost (#9540).

修复38

  • Fixed GitHub Copilot GPT models, including GPT-6 Astra, using the Chat Completions adapter instead of the required Responses adapter (#9253 by @petrroll).
  • Fixed inherited DeepSeek V4.1 thinking levels on OpenRouter and OpenCode Go preserving provider effort metadata (#9485).
  • Fixed inherited bodyless HTTP 400/413 errors from non-Cerebras providers being misclassified as context overflow (#9482).
  • Fixed inherited Vercel AI Gateway replaying unsigned thinking as assistant text (#9676).
  • Fixed inherited Google Generative AI and Vertex AI using unsupported thinking levels when reasoning is omitted or when model capabilities differ within a Gemini family (#9455).
  • Fixed inherited Anthropic-compatible relays breaking signed thinking replay when they report a different response model, while preserving fallback pricing (#9188).
  • Fixed inherited Amazon Bedrock one-hour cache writes being priced at the five-minute rate (#9457).
  • Fixed inherited quadratic CPU usage when draining buffered EventStream events (#9055).
  • Fixed inherited Mistral Medium reasoning requests to use reasoning_effort for all reasoning-capable mistral-medium-* model IDs instead of the unsupported prompt_mode (#8700).
  • Fixed inherited OpenCode and OpenCode Go requests to send x-opencode-session from sessionId across all supported API adapters (#9326).
  • Fixed inherited OpenAI Codex requests to send the model's Off reasoning effort instead of omitting it, while respecting unsupported Off mappings (#9191).
  • Fixed inherited Fireworks unsigned thinking replay and reasoning effort selection using catalog metadata, with verified DeepSeek V4 and Qwen3.8 fallbacks and removal of redundant GLM 5.2 and Kimi K3 effort aliases (#9323).
  • Fixed inherited OpenRouter requests to send x-session-id from sessionId for Chat Completions and Anthropic Messages models when prompt caching is enabled (#9102).
  • Fixed the inherited DeepSeek catalog to advertise deepseek-flash for DeepSeek V4.1 Flash instead of retired Flash aliases, and refreshed DeepSeek pricing metadata (#9423).
  • Fixed inherited Mistral-hosted GLM-5.2 reasoning requests to use reasoning_effort instead of the ignored prompt_mode (#9375).
  • Fixed inherited OpenAI-compatible Responses errors to identify the actual provider instead of always labeling them as OpenAI errors (#9298).
  • Fixed inherited Baseten requests to send session-affinity headers from sessionId for automatic prompt-cache routing (#9629).
  • Fixed inherited retry classification for Cloudflare 520 responses (#9627).
  • Fixed inherited retry classification for transient Azure peak-load capacity errors (#9669).
  • Fixed session tree navigation racing with active compaction and replacing its progress UI (#9179 by @acmerfight).
  • Fixed exact session ID lookup scanning complete transcript bodies instead of reading session headers (#9601 by @metaist).
  • Fixed repeated Anthropic thinking-drop notices being shown for the same dropped blocks, and shortened notices while retaining details in the session (#9391).
  • Fixed mid-run threshold compaction silently skipping oversized trailing tool results (#9740).
  • Fixed signal-terminated local shell commands being reported as successful with partial output (#9577 by @BrendanJMurphy).
  • Fixed local clipboard failures reporting success when the terminal ignored the fallback OSC 52 write, and added platform-specific setup guidance when no clipboard backend works (#9618).
  • Capped agent-level retry backoff at retry.maxAgentDelayMs (60s by default) so long retry runs stay responsive during prolonged transient outages (#8826).
  • Fixed direct RPC steer and follow_up commands bypassing extension input handlers (#8718).
  • Fixed premature missing-model errors after login by waiting for catalog discovery. Radius now defaults to balanced, falling back to the first available Radius model when needed.
  • Fixed fullscreen mode reserving a blank row for custom footers that render zero rows (#8919).
  • Fixed extension tools without parameter schemas to be rejected during registration instead of breaking provider requests (#9300).
  • Fixed before_agent_start handlers returning systemPrompt (and forceSystemPrompt) on models with mid-conversation system messages: the forced prompt is now sent as the provider's leading system prompt instead of being appended as a section patch after the original prompt.
  • Fixed loaded llama.cpp models with enable_thinking chat templates ignoring Pi's thinking level (#9528).
  • Fixed cancellation races that could start automatic compaction, leave stale retry state, or miss cancellation while waiting for summarization authentication (#9340, #9777).
  • Fixed asynchronous Kitty image conversion replacing newer partial tool output images (#8743 by @wutongyuonce).
  • Fixed inherited skill slash-command autocomplete ranking the skill: prefix instead of the bare skill name (#9120 by @yearth).
  • Fixed inherited file autocomplete boundaries and path quoting around CJK punctuation (#9746 by @haoqixu).
  • Fixed inherited LaTeX legacy font switches falling back to raw source, centered cases layouts around surrounding equations, and vertically laid out unsupported and nested display scripts (#8827, #9564, #7929).
  • Fixed inherited fullscreen Kitty images being erased by later row clears in WezTerm (#9169).

移除1

  • Removed unavailable inherited GPT-5.4 and GPT-5.4 mini models from OpenAI Codex selection (#9394).
0.85.1 2026-09-05 新功能 1新增 2修复 4 GPT-6 Astra

新功能1

  • GPT-6 Astra — Available through OpenAI API keys and OpenAI Codex subscriptions. See API Keys and OpenAI Codex.

新增2

  • Added GPT-6 Astra for OpenAI API keys and OpenAI Codex subscriptions.
  • Added five-times-faster mouse wheel scrolling while holding Alt in fullscreen mode (#9166 by @xl0).

修复4

  • Fixed configurable save keybindings in the model and thinking selectors (#9149 by @rwachtler).
  • Fixed SDK import failures caused by unintentionally publishing internal experimental code and dependencies in 0.85.0. The experimental client and experimental/plugin subpaths and server/client commands are now source-only through pi-test.sh; the supported local SDK and stdio RPC API are unchanged (#9132).
  • Fixed mouse hover changing selection and recentering autocomplete and settings lists, causing clicks to target a different item.
  • Fixed long prompt-cache requests for GPT-5.6+ Responses models to use prompt_cache_options.ttl: "30m" instead of prompt_cache_retention: "24h".
0.85.0 2026-09-04 新功能 3新增 5变更 2修复 25 Persistent Claude thinking effort

新功能3

  • Persistent Claude thinking effort — Supported Anthropic transports preserve per-turn effort and recover safely from signed-thinking mismatches. See Model Configuration.
  • Fullscreen transcript controls — Jump to the latest message from a scrolled transcript and use the embedded working indicator. See TUI Fullscreen Viewport.
  • Restorable in-memory sessions — Resume externally stored session entries through the SDK. See Session Management.

新增5

  • Added SessionManager.inMemory() support for restoring externally managed session entries (#8980 by @y-nk).
  • Added inherited OpenAI-compatible vllmPriority and supportsMaxOutputTokens model settings for vLLM scheduler priority and OpenAI Responses output-token limits (#9004 by @AppleDannyClegg, #8941 by @scturtle).
  • Added inherited LaTeX rendering for relational algebra join symbols (#9050 by @haoqixu).
  • Added a clickable "Jump to latest message" label with the tui.altScreen.bottom shortcut to the fullscreen transcript while it is scrolled up (#9080 by @rwachtler).
  • Added Meta (Muse subscription) login via /login meta with automatic Model API key refresh, plus META_API_KEY support.

变更2

  • Moved the streaming working indicator into the default editor border and matched its default spinner and label to the thinking-level border color. Custom editors retain the standalone indicator unless they opt in to embedding it (#8799 by @cristinaponcela).
  • Reduced inherited fullscreen transcript search latency on large transcripts by caching unchanged search results, indexing ASCII runs, and limiting highlight work to visible matches (#8800 by @cristinaponcela).

修复25

  • Fixed managed fd and ripgrep downloads on Linux musl systems (#9070 by @Charlie0113-T).
  • Removed the unavailable inherited Grok Build 0.1 model from /model (#9093 by @Jaaneek).
  • Fixed inherited provider streams emitting incompatible event sequences and custom tool-call deltas.
  • Restored the @earendil-works/pi-coding-agent/client compatibility entry point.
  • Fixed the inherited Qwen Token Plan Individual catalog to include Qwen3.8 Flash (#9021).
  • Fixed inherited OpenAI Codex SSE parsing to process terminal events that are not followed by a blank line (#9047).
  • Fixed inherited GitHub Copilot Claude Fable 5 requests so selected reasoning levels are sent (#8961).
  • Fixed inherited Baseten GLM-5.2 models incorrectly advertising image input support (#8293 by @Panoplos).
  • Fixed skills being unavailable when Bash is the only enabled tool (#8552 by @xl0).
  • Fixed concurrent session shares overwriting one another (#8613 by @wutongyuonce).
  • Fixed image orientation detection skipping EXIF data after non-EXIF APP1 segments (#8616 by @wutongyuonce).
  • Fixed imported sessions overwriting an existing session with the same filename (#8985 by @wutongyuonce).
  • Fixed session forks losing their compaction boundary (#8990 by @acmerfight).
  • Fixed in-memory session forks before an active turn settled (#8937 by @acmerfight).
  • Fixed inherited Fireworks GLM models using the wrong API adapter.
  • Fixed inherited NO_PROXY matching for root domains and subdomains (#8737 by @MeiSiristhebest).
  • Fixed bash, edit, find, grep, ls, read, and write tools ignoring ctx.cwd (#8627 by @vmizg).
  • Fixed inherited terminal startup under restricted seccomp policies that reject the SIGWINCH self-signal (#8898 by @bartlomiejkida).
  • Fixed inherited Zed terminal image capability detection (#8828 by @Perlence).
  • Fixed drag selection continuing over the fullscreen editor.
  • Fixed managed fd and ripgrep downloads requiring the GitHub Releases API (#8708 by @Terminator666666).
  • Fixed branch summaries failing when reasoning consumes the previous 2048-token output cap (#8845).
  • Fixed the write tool reporting UTF-16 code-unit counts as byte counts by removing the misleading count (#8979).
  • Fixed proxied plain-HTTP provider requests hanging after a tool call by tunneling them with CONNECT (#8134).
  • Fixed RPC abort reporting success without cancelling an in-progress manual compaction (#8920).
2026-085 个版本
0.84.4 2026-08-28 新功能 5新增 9变更 3修复 17 Terminal capability overrides

新功能5

  • Terminal capability overrides — Override detected terminal hyperlink, image, and truecolor support. See Capability Overrides.
  • Extension UI prompt events — Integrations can distinguish active agent work from time spent waiting for ctx.ui prompts. See Extension UI prompt events.
  • RPC queue clearing — Retrieve and clear queued steering and follow-up messages with clear_queue. See RPC C1.
  • Fullscreen selection copy controls — Disable automatic selection copying in fullscreen mode and use Ctrl+X to copy the active selection. See UI & Display.
  • DeepSeek V4 Flash Vision (experimental) — Use the vision-capable model through the built-in DeepSeek provider. See API Keys.

新增9

  • Added supportsMidConvoEffort to custom Anthropic Messages model compatibility settings.
  • Added transcript notices for Anthropic thinking blocks dropped during provider recovery when cache miss notices are enabled.
  • Added ui_prompt_start and ui_prompt_end extension events so host integrations can distinguish active agent work from waiting on user-facing ctx.ui prompts (#8355 by @cristinaponcela).
  • Added detectSupportedImageMimeTypeFromFile() to the public library exports (#8600 by @xl0).
  • Added inherited experimental vision-capable deepseek-v4-flash-vision-exp model support.
  • Added transcript usage notices for compaction and branch summaries when cache miss notices are enabled.
  • Added RPC clear_queue to retrieve and remove queued steering and follow-up messages (#8432).
  • Added environment variables and advanced settings for overriding auto-detected terminal hyperlink, image, and truecolor capabilities (#8665).
  • Added fullscreenCopyOnSelect to disable automatic fullscreen selection copy; when disabled, Ctrl+X copies the active text selection before falling back to the last assistant message, while /tree still copies the selected message (#7720).

变更3

  • Changed fullscreen scrollbars to reveal on pointer entry, support optional scrollbarTrack and scrollbarThumb theme colors with muted and text fallbacks, keep one thumb color across normal and expanded states, and support track-click jumping.
  • Changed fullscreen transcript search arrows to underline on hover and capitalized the search placeholder.
  • Changed selectors in /thinking, /model, /scoped-models, /trust, per-model thinking settings, and theme settings to keep active options marked while browsing. /scoped-models now uses consistent per-item toggles and strikes through unavailable models (#8900).

修复17

  • Fixed toggling thinking visibility clearing partial output from running Bash tools (#8611).
  • Fixed Windows shell aborts crashing Pi when taskkill.exe is unavailable on PATH (#6596).
  • Fixed resumed sessions corrupting the next appended entry when their JSONL file lacks a trailing newline (#8345).
  • Fixed extension messages sent with triggerTurn: false while the agent is running being inserted between a tool call and its result, which made providers that validate message order reject the replayed history. They are now appended once the turn's tool results are in (#8537).
  • Fixed compaction and branch summaries forcing toolChoice: "none" (#8649, #8638).
  • Fixed large tool results crossing the auto-compaction threshold being sent to the provider before compaction. Pi now compacts between tool execution and the next assistant response in the same run, and restores interactive progress when that run resumes (#6879).
  • Fixed Google Vertex requests failing with HttpsProxyAgent is not a constructor when the bundled Node.js runtime uses an HTTP(S) proxy (#8610).
  • Fixed saving a default model from a non-empty model scope so it remains available in that scope.
  • Fixed inherited @ file autocomplete ranking to prefer direct and shallower matches over similarly ranked nested paths (#8669).
  • Fixed inherited OpenAI-compatible streams serializing thinking signatures repeatedly during streaming (#8671).
  • Fixed inherited main-screen rendering crashing when image-heavy output exceeded V8's string length limit (#8028).
  • Fixed inherited fullscreen double-click word selection splitting paths and kebab-case tokens on / and - (#8676).
  • Fixed inherited Cloudflare AI Gateway catalogs omitting supported workers-ai/* passthrough models.
  • Fixed inherited OpenAI-compatible reasoning replay to merge consecutive streamed text and summary reasoning_details deltas.
  • Fixed inherited OpenRouter reasoning controls so reasoning-mandatory models do not receive effort: "none" (#8614 by @davidbrai).
  • Fixed inherited OpenAI-compatible Chat Completions ignoring an explicitly requested toolChoice when no tools are defined.
  • Fixed inherited fragmented Mistral tool calls splitting when continuation chunks omit the tool-call ID (#8387).
0.84.3 2026-08-24 新功能 3破坏性变更 1新增 10变更 10修复 40 PowerShell tool

新功能3

  • PowerShell tool — Use optional native PowerShell command execution on Windows. See PowerShell Tool.
  • Safer managed updates — Stage, verify, and atomically activate updates for installer-managed installations. See Install and Manage.
  • Model and thinking controls — Select thinking levels with /thinking, search defaults, keep selections session-scoped, and persist them explicitly with Ctrl+S. See Models and Thinking.

破坏性变更1

  • Renamed the inherited GoogleThinkingLevel type to GoogleApiThinkingLevel and added ResolvedGoogleThinkingLevel for normalized adapter levels.

新增10

  • Added an optional powershell tool for Windows, configurable through defaultTools and the SDK. See PowerShell Tool.
  • Added a /thinking selector and searchable default choices to the model and thinking selectors; Ctrl+S saves the selected model as the global default. See Models and Thinking.
  • Added optional routing session IDs to exported compaction summary helpers so callers can preserve provider routing without enabling prompt cache writes.
  • Added transcript usage notices for compaction and branch summaries when cache miss notices are enabled.
  • Added session_compact_failed extension events so compaction failures and aborts expose their reason, retry state, source, and error message to handlers (#8175).
  • Added inherited provider-neutral toolChoice support to simple stream requests.
  • Added inherited automatic Anthropic server-side refusal fallback for supported first-party models, including returned-model usage pricing (#8017).
  • Added inherited configurable OpenAI-compatible thinking-token budget fields for vLLM, Qwen/SGLang, and llama.cpp servers. See OpenAI Compatibility (#8275 by @bnsd55).
  • Added inherited China-specific ZAI Coding Plan models, including GLM-4.6V vision support and API-equivalent usage cost estimates (#8220).
  • Added inherited deepseek-v4-pro-0813 support to the Qwen Token Plan Individual catalog (#8194).

变更10

  • Changed experimental installer-managed installations so pi update stages, verifies, and atomically activates the selected release in place. See Install and Manage.
  • Changed inherited built-in xAI models to use the Responses API with encrypted reasoning replay and made Grok 4.6 the default xAI model (#8124 by @Jaaneek).
  • Changed inherited Anthropic, Azure OpenAI, Google, Mistral, and OpenAI adapters to send Pi's default User-Agent unless overridden (#8305).
  • Changed Windows and WSL keybinding defaults to avoid terminal-reserved shortcuts for image paste, model cycling, editor undo, fullscreen transcript navigation and search, and message queueing (#8372).
  • Changed Bun release archives to ship the native clipboard binary only inside the wrapper package, removing a duplicate platform package from each archive.
  • Changed package resource glob expansion to use Node.js's built-in implementation with deterministic visible-path matching, reducing the installed runtime dependency tree.
  • Changed the bundled Node.js runtime to load jiti only when importing an extension and Babel only when uncached source needs transformation, reducing CLI startup time and bundle size.
  • Changed syntax highlighting to initialize only twenty common languages eagerly and defer the remaining grammars until after the initial TUI render, reducing CLI startup time.
  • Changed the Node.js CLI and RPC entrypoints to load a bundled runtime, reducing startup filesystem reads while keeping the public library and legacy module paths on the modular runtime for normal dependency identity.
  • Changed session sharing to render clickable terminal links, display only the canonical Radius artifact URL, and include the current system prompt and active tool definitions in Radius session shares.

修复40

  • Fixed failed extension factories leaving event subscriptions, provider registrations, and default flag state active (#8424 by @acmerfight).
  • Fixed models.json typings omitting the documented OpenAI-compatible compat.supportsFinishReason provider and model override (#8487 by @petrroll).
  • Fixed /model and /thinking selections being persisted globally unless explicitly saved with Ctrl+S (#5263).
  • Fixed JSON and RPC toolcall_start events omitting the tool call id and name (#7953 by @christianklotz).
  • Fixed extensions failing to load when the Node.js CLI runs as a single-executable application (#8237).
  • Fixed nested Markdown skills inside .agents/skills/ grouping directories not being discovered.
  • Fixed compaction and branch summarization requests exposing tools to providers.
  • Fixed single-object edit tool inputs failing validation by accepting them as one-edit arrays in both coding-agent and harness edit tools (#7835).
  • Fixed root Markdown files such as README.md and AGENTS.md in skill directories being reported as broken skills unless they declare valid skill frontmatter (#7805).
  • Fixed the default Cerebras model referencing an unavailable Z.AI model.
  • Fixed inherited OpenAI-compatible Chat Completions reasoning replay to preserve and resend assistant-level reasoning_details verbatim and in order (#7994).
  • Fixed inherited Anthropic server-side fallback responses being priced with the requested model instead of the returned fallback model (#8285).
  • Fixed inherited GitHub Copilot login triggering model-policy rate limits by limiting policy updates, retrying model discovery once, and honoring server retry delays (#7850).
  • Fixed inherited Amazon Bedrock dropping and failing to replay opaque redacted reasoning from non-Anthropic models (#8314 by @seiji).
  • Fixed inherited Z.AI Coding Plan models deriving incomplete reasoning-effort metadata, including missing GLM-5.3 low, high, and max levels (#8336).
  • Fixed inherited DeepSeek V4 Flash on OpenCode and OpenCode Go omitting its supported low thinking level (#8181 by @tianshuang).
  • Fixed inherited Azure OpenAI Responses ignoring toolChoice in provider-specific stream requests.
  • Fixed inherited Amazon Bedrock response hooks receiving only a synthesized request id instead of the raw response headers (#8234).
  • Fixed inherited Kimi usage reporting so top-level cached_tokens count as cache reads instead of normal input tokens (#8075).
  • Fixed inherited Google custom models ignoring thinkingLevelMap, which dropped extended thinking controls (#8135).
  • Fixed writes to auth.json and models-store.json overriding administrator-managed file permissions and ACLs (#7779).
  • Fixed UTF-8 BOM markers preventing frontmatter and user configuration files from loading (#8337).
  • Fixed invalid settings files being easy to miss during interactive startup by rendering warnings with the file path inside the TUI (#7829).
  • Fixed the subagent example repeatedly prompting before running project-local agents in trusted repositories (#8261).
  • Added session_compact_failed extension events so compaction failures and aborts expose their reason, retry state, source, and error message to handlers (#8175).
  • Fixed truncated compaction and branch summaries being persisted when generation reaches its output token limit (#7048).
  • Fixed npm package update checks treating older registry versions as available updates, preventing pi update from downgrading already-newer installed packages (#8226).
  • Fixed built-in llama.cpp models disappearing from /model when /llama refreshed a configured server under PI_OFFLINE, and included idle-slept sleeping router models plus autoloadable unloaded presets in the selectable catalog (#8558 by @cristinaponcela).
  • Fixed pi.registerFlag() accepting default values that do not match the declared flag type (#8064).
  • Fixed Z.AI Coding Plan defaults referencing the removed GLM-5.1 model (#8096).
  • Fixed repeated ambiguous truncated-response recovery being mislabeled as context overflow (#8130).
  • Fixed duplicate fullscreen right-click paste in VS Code-based terminals on Windows (#8186).
  • Fixed inherited padded text exceeding narrow terminal widths (#8252).
  • Fixed inherited wrapped Markdown table links leaking color into borders and neighboring cells, including tables inside blockquotes (#8335).
  • Fixed llama.cpp login guidance to direct users to /llama before /model when no local models are loaded (#8203).
  • Fixed hung pi.dev model catalog requests consuming the entire refresh deadline without retrying (#8198).
  • Fixed inherited Xiaomi model catalogs listing shut-down MiMo V2 models in /model and --list-models (#8187).
  • Fixed branch summary entries recording the navigation destination in fromId instead of the pre-navigation source leaf.
  • Fixed threshold auto-compaction being skipped when providers omit streaming usage data (#8328).
  • Fixed dash-prefixed prompts being parsed as options by supporting -- as an end-of-options delimiter (#7269).
0.84.2 2026-08-14 新功能 3新增 10变更 5修复 27 Fullscreen transcript search

新功能3

  • Fullscreen transcript search — Search and navigate matches in fullscreen mode. See TUI Fullscreen Viewport.
  • Configurable default tools — Choose startup built-in tools globally or per project. See Tools.
  • Configurable fullscreen exit output — Print the transcript or only a resume hint on exit. See Interactive Mode.

新增10

  • Added per-block fullscreen mouse expansion for thinking sections and tool results, while preserving drag selection and link activation.
  • Added fullscreen transcript search with Ctrl+Shift+F, incremental match highlighting, configurable search match theme colors, and next/previous navigation with Enter/Ctrl+G and Shift+Enter/Ctrl+Shift+G.
  • Added experimental strict JSON-schema constrained sampling for the default read, bash, edit, and write tools under PI_EXPERIMENTAL=1.
  • Added a fullscreen exit output setting to choose between printing the final transcript and only a session resume hint.
  • Added the defaultTools setting for configuring the initial built-in tool selection globally or per project.
  • Added --use-theme <name[/name]> to choose an initial per-run interactive theme without changing saved settings (#7722 by @rwachtler).
  • Added expandPromptTemplates to extension pi.sendUserMessage() options for explicitly dispatching commands and expanding skills and prompt templates. See C2 (#7857 by @mrexodia).
  • Added inherited createGatewayBindingFetch() for routing Cloudflare AI Gateway requests through a Workers AI binding without an API token (#7901 by @Maximo-Guk).
  • Added inherited AssistantMessage.endTurn to preserve OpenAI Codex's terminal end_turn signal for diagnostics (#7766).
  • Added inherited unbound single-line transcript scrolling actions for fullscreen mode. See TUI Fullscreen Viewport (#7903 by @midastruth).

变更5

  • Changed inherited Kimi Coding requests to use pi's runtime User-Agent header.
  • Replaced the inherited Mistral SDK transport with a native Chat Completions HTTP stream, eliminating its generated client and schema runtime overhead.
  • Documented the generic AI_AGENT=pi process marker and how it differs from PI_CODING_AGENT=true (#7747).
  • Changed inherited OpenAI Responses deferred tool loading to prefer message-anchored additional_tools where supported while retaining tool-search and top-level fallbacks (#7709).
  • Reduced inherited fullscreen rendering allocation churn by painting full-width layout rows directly instead of recompositing them on every frame.

修复27

  • Fixed root Markdown files such as README.md and AGENTS.md in skill directories being reported as broken skills unless they declare valid skill frontmatter (#7805).
  • Fixed single-object edit tool inputs failing validation by accepting them as one-edit arrays in both coding-agent and harness edit tools (#7835).
  • Fixed managed-tool downloads delaying TUI startup and hiding diagnostics in fullscreen mode by mounting the TUI first and showing download progress and warnings inside it.
  • Fixed opening a model selector immediately after startup cancelling and restarting the in-progress model catalog refresh.
  • Fixed inherited GitHub Copilot login triggering API rate limits while enabling model policies by limiting concurrent policy updates (#6187).
  • Fixed fullscreen transcript search snapping back to the current match during manual scrolling and fragmented mouse input leaking into the search query.
  • Fixed inherited required LaTeX arguments starting on a new line being parsed as empty (#7760).
  • Updated the transitive nanoid development dependency to address a denial-of-service vulnerability.
  • Fixed fallback rendering for extension tool results to collapse long output and honor tool expansion (#7979).
  • Fixed JSON and RPC message_update events dropping cumulative usage during streaming. See JSON Event Mode and RPC C1 (#7982 by @christianklotz).
  • Fixed pi.sendMessage(..., { triggerTurn: false }) steering an active run instead of only recording the custom message (#8022 by @cristinaponcela).
  • Fixed the defaultTools setting dropping extension and SDK custom tools when selecting built-in defaults.
  • Fixed the subagent example rejecting YAML array syntax for the tools frontmatter field (#7598 by @alexsavio).
  • Fixed the subagent example dropping parent session model, thinking, and tool configuration (#7897 by @virtuald).
  • Fixed custom system prompts concatenating the current working directory with later appended prompt content (#7887 by @distributedlock).
  • Fixed inherited OpenAI Responses function and custom tool calls losing namespaces during streaming, proxying, and replay (#7709).
  • Fixed inherited upstream request buffer failures not triggering automatic assistant retries.
  • Fixed inherited built-in and custom DeepSeek API models sending output limits through an unsupported field.
  • Fixed inherited Amazon Bedrock replay rejecting tool arguments that contain empty object keys while preserving all valid nested values (#7882 by @muyiyr).
  • Fixed inherited DeepSeek compatibility detection for base URLs whose hostname contains uppercase letters (#7933 by @yearth).
  • Fixed inherited Google Generative AI and Vertex AI responses with tool calls incorrectly treating output-limit or provider-error stops as normal tool use (#8059).
  • Fixed inherited fullscreen mouse drag selection and OSC 8 link activation in terminals that report generic SGR mouse release button codes (#7963).
  • Fixed inherited focused fullscreen overlays not receiving mouse wheel or viewport scroll keys such as PageUp and PageDown (#7894).
  • Fixed inherited LaTeX control spaces split across line endings causing complete expressions to fall back to raw source.
  • Fixed split Alt+Enter input over SSH being misread as Escape, added PI_TUI_ESC_TIMEOUT for high-latency terminals, and limited that timeout to lone Escape input (#7899 by @powerfooI).
  • Fixed inherited idle fullscreen sessions repainting and clearing text selection when the terminal loses focus (#7892 by @terrorobe).
  • Fixed fullscreen selection copy to use the host clipboard and report failure instead of claiming success when OSC 52 is unsupported (#8110 by @Panoplos).
0.84.1 2026-08-07 新功能 4新增 5变更 2修复 6 Qwen Token Plan Individual

新功能4

  • Qwen Token Plan Individual — Use the built-in provider for models documented for Individual subscriptions. See API Keys.
  • Authentication readiness checks — Use pi auth check to verify provider or model credentials, optionally emitting the resolved credential.
  • Improved fullscreen interaction — Select words and paragraphs with multiple clicks and configure half-page transcript scrolling. See TUI Fullscreen Viewport.
  • Terminating blocked tool calls — Extension tool_call handlers can stop all-terminating batches without another model call. See Tool Events.

新增5

  • Added Qwen Token Plan Individual as a built-in provider with its documented subscription model catalog and the shared international QWEN_TOKEN_PLAN_API_KEY. See API Keys (#7659 by @arasovic).
  • Added pi auth check provider/model auth preflight with optional credential output (#7152).
  • Added terminate support to blocked extension tool_call events so all-terminating batches can skip the automatic follow-up model call. See Tool Events (#7715 by @muyiyr).
  • Added inherited double-click word and whitespace selection, granularity-aware drag selection, and triple-click paragraph selection in fullscreen mode (#7725, #7733 by @volsa).
  • Added inherited unbound half-page transcript scrolling actions for fullscreen mode. See TUI Fullscreen Viewport (#7735).

变更2

  • Softened the bash tool's PI_* environment guideline in an attempt to reduce unnecessary inspection commands (#7128).
  • Reduced worst-case automatic terminal theme detection delay from 200 ms to 100 ms by probing color-scheme and background support concurrently.

修复6

  • Fixed Bun standalone binaries crashing on startup when the cwd contains a bunfig.toml with preload by compiling with --no-compile-autoload-bunfig (#7685 by @geril07).
  • Fixed extension TUI method wrappers recursing indefinitely when delegating to the original method (#7731).
  • Fixed right-click not pasting clipboard text in fullscreen mode on Windows.
  • Fixed inherited Agent.reset() clearing transcript and runtime state during active runs; it now rejects until the agent is idle (#7717 by @wesleyzhangwq).
  • Fixed inherited LaTeX relation, multiplication, and named-operator spacing, and matrix composition with stacked fractions, operator limits, and adjacent matrices.
  • Reduced inherited fullscreen mouse event volume under tmux, Zellij, and GNU Screen by using button-motion tracking instead of all-motion tracking.
0.84.0 2026-08-06 新功能 5破坏性变更 12新增 26变更 2修复 74 Fullscreen TUI mode

新功能5

  • Fullscreen TUI mode — Switch between regular and fullscreen modes at runtime, with a sticky editor and footer, independently scrollable transcript, and draggable scrollbars. See UI & Display.
  • Mermaid and LaTeX rendering — Render Mermaid diagrams and terminal-friendly Unicode math in interactive transcripts. See Markdown settings and TUI Markdown.
  • Per-directory context overrides — Use AGENTS.override.md to replace context files for a specific directory. See Context Files.
  • Advanced custom model sampling — Configure arbitrary OpenAI-compatible samplingParams and opt-in vLLM thinking_token_budget values. See Sampling Parameters.
  • Baseten provider — Use built-in Baseten authentication and model support. See API Keys.

破坏性变更12

  • Renamed the inherited pi-ai ModelsStreamTransforms interface to ModelsRequestTransforms because its header transformation now applies to all authenticated provider requests.
  • Changed JSON and RPC message_update events to emit only assistantMessageEvent deltas, removing the cumulative message and assistantMessageEvent.partial fields that caused quadratic output growth. Clients that need partial messages must assemble deltas between message_start and message_end; the latter remains authoritative (#7290).
  • ModelRegistry.getApiKeyAndHeaders() now returns ProviderHeaders with string | null values and preserves null header-deletion markers. Extensions that inspect returned headers must handle null; extensions forwarding them to pi-ai streams should pass them through unchanged. This prevents placeholder OpenAI credentials from being sent through Cloudflare AI Gateway (#7030).
  • Changed ModelRegistry.refresh() to accept ModelsRefreshOptions and return ModelsRefreshResult instead of discarding cancellation and provider errors.
  • Changed ModelRuntime.setRuntimeApiKey() to accept auth cancellation options rather than catalog refresh options. Call refresh({ providers: [providerId], signal }) separately when remote freshness is required.
  • Required config-form extension OAuth refreshToken(credentials, signal) callbacks to accept and honor a concrete abort signal.
  • Replaced dynamic provider refresh context store access with the read-only context.stored snapshot and generation-checked context.publish() transaction.

    Providers built with createProvider({ fetchModels }): no catalog-publication migration is required. Before and after, return the fetched models and register the resulting provider; createProvider() owns restoration, persistence, and in-memory publication.

    // Before
const beforeProvider = createProvider({
  // ...
  fetchModels: async ({ signal }) => {
    const response = await fetch(catalogUrl, { signal });
    return parseModels(await response.json());
  },
});
pi.registerProvider(beforeProvider);
// After: unchanged
const afterProvider = createProvider({
  // ...
  fetchModels: async ({ signal }) => {
    const response = await fetch(catalogUrl, { signal });
    return parseModels(await response.json());
  },
});
pi.registerProvider(afterProvider);

    Handwritten native Provider.refreshModels(): replace direct store access and pre-publication mutation with generation-guarded publications.

    // Before
refreshModels: async (context) => {
  const stored = await context.store.read();
  if (stored) currentModels = stored.models;
  if (!context.allowNetwork) return;
  const refreshed = await fetchModels(context.signal);
  currentModels = refreshed;
  await context.store.write({ models: refreshed, checkedAt: Date.now() });
},
// After
refreshModels: async (context) => {
  if (context.stored) {
    const restored = context.stored.models;
    if (!(await context.publish({
      update: () => { currentModels = restored; },
    }))) return;
  }
  if (!context.allowNetwork) return;
  const refreshed = await fetchModels(context.signal);
  if (context.signal.aborted) return;
  await context.publish({
    persist: { models: refreshed, checkedAt: Date.now() },
    update: () => { currentModels = refreshed; },
  });
},

    For the config-form pi.registerProvider(name, { refreshModels }), callbacks that only return models remain unchanged; pi publishes the returned list. If such a callback previously used context.store for custom persistence, read context.stored and call context.publish({ persist: entry }). In publish(), omit persist to leave storage unchanged, pass a ModelsStoreEntry to write it, or pass persist: null to delete it.

  • Replaced the inherited pi-agent-core harness session model with the v4 lane-based Session, SessionStorage, and SessionRepo APIs, including durable operation records, global facts, shared sequence numbers, and tree-scoped lane views.
  • Promoted the inherited v2 session and AgentHarness API from pi-agent-core's experimental entrypoint to its default export and removed the experimental subpaths.
  • Removed the inherited legacy JSONL and in-memory repository APIs. Use pi-agent-core's v4 JsonlSessionRepo or InMemorySessionRepo, both implementing the new SessionRepo contract.
  • Added the inherited required pi-agent-core FileSystem.renameFile() operation for atomic JSONL publication; custom harness file-system implementations must provide same-filesystem replacement semantics (#7707 by @davidbrai).
  • Replaced experimental remote-session list summaries with durable SessionMetadata; RemoteSession.sessions no longer exposes runtime phase, model, thinking, attachment, or lock state, which remains available from acquired SessionSnapshot values (#7708).

新增26

  • Added built-in Baseten provider support with BASETEN_API_KEY authentication and zai-org/GLM-5.2 as the default model.
  • Added experimental remote-session client APIs: the transport-neutral PiClient, CBOR protocol, Unix-socket transport, and @earendil-works/pi-coding-agent/client RemoteSession controller with transcript reducers. See Pi Client and Remote Protocol (#7344, #7348, #7371, #7409).
  • Added CredentialSynchronizationError for credential changes that commit successfully but fail to synchronize local model state.
  • Added chainable pi.registerMarkdownTransformer() hooks for display-only transformation of user and assistant Markdown. See C1 (#7231 by @xl0).
  • Added an experimental fullscreen TUI mode, selectable through --tui-mode fullscreen or /settings (#7304).
  • Added runtime switching between regular and fullscreen TUI modes through /settings.
  • Added a sticky editor, status, widget, and footer dock to fullscreen mode while keeping the transcript independently scrollable.
  • Added a draggable transcript scrollbar to fullscreen mode with configurable auto, always, and hidden modes through /settings; always reserves the rightmost column.
  • Added page scrolling and marked-message navigation shortcuts to fullscreen mode.
  • Added an optional scrollbarThumb theme color for fullscreen scrollbar thumbs, falling back to selectedBg.
  • Added configurable themed Unicode rendering for supported Mermaid diagrams in interactive messages, including optional rendering while streaming. See Markdown settings (#7624 by @xl0).
  • Added opt-in Ctrl+P/Ctrl+N prompt history navigation, with explicit history bindings taking precedence over application shortcuts while the editor is focused.
  • Added per-directory AGENTS.override.md context files, which replace AGENTS.md or CLAUDE.md in the same directory while preserving context from other directories. See Context Files (#7681 by @Marvae).
  • Added AI_AGENT=pi to CLI and RPC child-process environments for generic agent attribution. See Environment Variables (#7493 by @renaudhartert-db).
  • Added inherited terminal-friendly Unicode rendering for LaTeX expressions in Markdown. See TUI Markdown.
  • Added stacked transient notifications in fullscreen mode.
  • Added arbitrary OpenAI-compatible model sampling parameters through samplingParams in models.json, model overrides, extension providers, and stream options. See Sampling Parameters (#7568 by @mrexodia).
  • Added inherited opt-in vLLM thinking_token_budget support for OpenAI-compatible models, reserving output tokens for the final answer (#7638 by @bnsd55).
  • Added inherited support for OpenAI-compatible streams that omit finish_reason, using compat.supportsFinishReason to infer normal and tool-use stops when the stream ends. See OpenAI Compatibility.
  • Added inherited deferred provider request contracts, durable response handles, authenticated fetch/cancel dispatch, and faux-provider support for pending, ready, failed, and cancelled responses (#7339 by @davidbrai).
  • Added inherited vendor-neutral telemetry contracts plus agent-owned typed AI-request and harness schemas, composed span starters, and callback helpers. See the agent telemetry schema reference.
  • Added inherited structured Amazon Bedrock failure diagnostics with HTTP status, modeled error code, and AWS request id when available (#7286 by @brianstanley).
  • Added inherited AgentOptions.shouldStopAfterTurn for gracefully stopping after a completed turn before queued messages or another model call are processed. See Agent Options (#7367 by @acmerfight).
  • Added inherited v4 JsonlSessionRepo support for append-only JSONL harness sessions (#7611 by @davidbrai).
  • Added inherited bounded branch-entry and indexed open-operation recovery queries to the v4 session API (#7448, #7646).
  • Added the inherited compile-complete AgentHarness v2 scaffold; unfinished operation paths reject with HarnessNotImplemented while durable execution is implemented.

变更2

  • Added inherited optional cancellation to pi-ai ModelsStore reads, writes, and deletions; catalog orchestration binds these waits to the provider refresh signal.
  • Reduced the inherited default fullscreen mouse wheel step from three lines to one for finer scrolling.

修复74

  • Fixed the footer showing (sub) for generic OAuth/OpenID sign-ins without a known subscription; extension OAuth providers can opt in with isSubscription.
  • Fixed inherited OAuth token refreshes so stalled requests release the credential-store lock (#7508).
  • Fixed inherited tool argument validation to preserve values that already match an anyOf/oneOf union arm before coercion, avoiding nullable unions converting null to another primitive value (#7328).
  • Fixed inherited Fireworks GLM 5.2 requests sending the unsupported prompt_cache_retention field when long cache retention is enabled, and enabled session affinity for automatic prompt caching (#7676).
  • Fixed inherited JsonlSessionRepo enforcing session IDs globally across working directories; IDs are now unique within each working directory.
  • Fixed inherited JSONL session forks and torn-tail repairs to publish atomically, avoiding partially written or corrupted sessions after interrupted writes (#7707 by @davidbrai).
  • Fixed path-containing find globs returning no results on Windows (#6817).
  • Fixed messages queued during manual /compact failing instead of being sent after compaction completes.
  • Fixed Git Bash, MSYS, Cygwin, and WSL drive paths passed to built-in file tools resolving against the current Windows drive instead of their native drive (#7064, #7547).
  • Fixed project-level nested provider retry settings replacing unmodified global provider retry settings (#7572).
  • Fixed inherited GitHub Copilot Grok 4.5 requests to use the supported Responses API (#7560).
  • Fixed fullscreen shutdown leaking terminal capability-query replies into the parent shell prompt.
  • Fixed bare exact --model IDs shared by multiple providers choosing the first catalog entry instead of the sole authenticated provider or a clear ambiguity error (#7327).
  • Fixed standalone x64 binaries requiring Haswell-era AVX2/BMI2 instructions by compiling release executables against Bun's baseline runtime (#7390 by @davidbrai).
  • Fixed Ctrl+X copy confirmations in fullscreen mode adding a transcript status line instead of showing the transient Copied! marker.
  • Fixed Kitty image previews in fullscreen mode overlapping the sticky editor and footer dock while scrolling.
  • Fixed image-heavy fullscreen sessions lagging when layout changes retransmitted visible Kitty image payloads and rendered the transcript twice per frame.
  • Fixed spaces in /settings searches toggling the highlighted setting while typing multi-word queries such as TUI mode or Quiet startup.
  • Fixed custom editors not inheriting the default editor's autocomplete dropdown item limit (#7333).
  • Fixed malformed resource arrays in package manifests crashing session startup (#7187).
  • Fixed the DOOM overlay example downloading its shareware WAD from a dead URL.
  • Fixed setToolsExpanded(false) to be a no-op when tool output is already collapsed, avoiding redundant Tool output: collapsed startup notices from extensions (#7292).
  • Fixed extension-driven model calls in custom compaction, handoff, and Q&A examples to dispatch through the coding-agent model runtime so custom providers and resolved auth options are preserved (#7325).
  • Fixed long-running sessions using stale credentials after another process updates auth.json without serializing concurrent credential reads and delaying startup (#7319).
  • Fixed concurrent models-store.json reads forming a file-lock convoy and delaying startup.
  • Updated the packaged brace-expansion dependency to 5.0.8 to address GHSA-mh99-v99m-4gvg (#7316).
  • Fixed forced model availability refreshes remaining blocked behind a stalled earlier refresh (#7301, #7421 by @a-yeyang).
  • Fixed /model catalog refresh failures to identify every catalog that failed.
  • Fixed provider login remaining stuck after saving credentials when a model catalog refresh stalls by separating local credential consistency from bounded background freshness (#7027, #7113, #7418).
  • Fixed /scoped-models waiting for remote catalogs before rendering instead of showing cached models and cancelling refresh on close (#7153).
  • Fixed /model <name> waiting for catalog refresh before checking cached model matches (#7443).
  • Fixed stale availability snapshots and errors publishing after a newer availability pass.
  • Fixed stale pi.dev, Radius, llama.cpp, and extension catalog refreshes publishing after a newer provider refresh.
  • Fixed cancellation while waiting for file-backed credential or model-catalog locks, preventing cancelled mutations from running or committing later.
  • Fixed concurrent in-memory credential mutations losing unrelated provider updates by serializing their read-modify-write sections.
  • Updated undici to 8.9.0 and the packaged brace-expansion to 5.0.9 to address GHSA-8xcm-r25x-g524, GHSA-4cwx-7wf7-3272, GHSA-m8rv-5g2x-5cg5, GHSA-jr45-8vmc-qm54, GHSA-v3r7-h72x-cjcm, and GHSA-rgw5-rvv9-x895.
  • Fixed GitHub Copilot compaction and branch summaries using the Individual endpoint instead of the credential-resolved Business or Enterprise endpoint (#6768).
  • Fixed extension model calls dropping credential-resolved endpoints when forwarding request authentication, including custom compaction with GitHub Copilot Business and Enterprise accounts (#7579).
  • Fixed fullscreen transcript navigation leaving no editor-accessible Home, End, PageUp, or PageDown variants by adding Ctrl-modified editor bindings (#7574).
  • Fixed extension event-bus listeners surviving session reloads and disposal (#7656 by @tudoroancea).
  • Fixed /copy failing to read clipboard text on Wayland when no X11 clipboard is available (#7387).
  • Fixed slow connections failing during the initial connection attempt by increasing the connect timeout (#7435 by @muyiyr).
  • Fixed oversized images returned by extension and built-in tools bypassing automatic image resizing. See Image settings (#7330 by @tizmagik).
  • Fixed session discovery missing sessions stored through symlinked directories (#7552 by @muyiyr).
  • Fixed manual compaction racing with threshold auto-compaction (#7370 by @davidbrai).
  • Fixed responses truncated below their intended output limit ending the run instead of compacting and retrying once (#7540 by @davidbrai).
  • Fixed Git package updates leaving dependencies missing when git clean cannot remove an ignored dependency directory (#7570 by @mrexodia).
  • Fixed find results from POSIX and Windows filesystem roots losing the first path segment or gaining duplicate trailing separators (#7569 by @petrroll).
  • Fixed transient version-check, catalog, managed-tool, and package-management HTTP failures not being retried (#7632 by @petrroll).
  • Fixed interactive errors ignoring the configured output padding.
  • Fixed the inherited OpenCode Go provider display name.
  • Fixed inherited provider error normalization treating arrays and class instances as structured response bodies instead of preserving their original errors (#7205 by @erikogenvik).
  • Fixed inherited Anthropic streams dropping text or thinking included in the initial content-block event (#7358 by @davidbrai).
  • Fixed inherited Google history conversion dropping signed empty text and thinking blocks required for replay (#7362 by @jingtao-wisdomgraph).
  • Fixed inherited OpenAI Codex cached WebSocket sessions being shared across different account credentials (#7364).
  • Fixed inherited transient Google Generative AI and Vertex AI provider errors bypassing automatic retries (#7471 by @vish-pr).
  • Fixed inherited Gemini 3 tool call ids being discarded during history conversion, breaking signed multi-turn replay (#7494 by @muyiyr).
  • Restored inherited GitHub Copilot models returned through account-specific policy responses (#7672 by @muyiyr).
  • Replaced the inherited retired Qwen Token Plan qwen3.8-max-preview model with qwen3.8-max (#7670 by @QuintinShaw).
  • Fixed inherited terminal width accounting for Indic conjunct grapheme clusters (#6987 by @petrroll).
  • Fixed inherited nested fullscreen stack layouts ignoring child minimum sizes.
  • Fixed inherited batched terminal color-scheme reports being parsed as one malformed response (#7550).
  • Fixed inherited terminal progress clearing to emit the complete OSC 9;4 sequence (#7581).
  • Fixed inherited iTerm2 image payloads omitting the size metadata required by the xterm.js image addon (#7612).
  • Fixed inherited width truncation leaving OSC 8 hyperlinks unterminated (#7657 by @xXJSONDeruloXx).
  • Updated inherited GPT-5.6 Terra and Luna pricing across OpenAI and passthrough model catalogs.
  • Fixed inherited Fireworks Kimi K3 models to use the OpenAI-compatible API with native reasoning-effort levels and deferred tools (#7199, #7230 by @XBeg9).
  • Updated the inherited Groq Qwen reasoning override for the replacement qwen/qwen3.6-27b model.
  • Fixed inherited Windows Shift+Enter detection by reading modifier state from the native Win32 helper.
  • Fixed the inherited pi-tui npm package omitting the source and build scripts needed to rebuild its Windows and Darwin native addons.
  • Fixed inherited Windows console truecolor detection when Windows Terminal does not provide WT_SESSION to child shells.
  • Fixed inherited phantom fullscreen text selection from unmatched mouse events when changing terminal pane focus.
  • Fixed inherited keyboard input rendering latency on Windows by letting input preempt the throttled render timer.
  • Fixed inherited agent harness path handling on Windows for file basenames, recursive skill loading, and prompt template names.
2026-0712 个版本
0.83.0 2026-07-29 新功能 3破坏性变更 1新增 7变更 1修复 16 Credential export for external clients

新功能3

  • Credential export for external clients — pi auth print-api-key and pi auth print-bearer-token export configured credentials with automatic OAuth refresh and minimum-validity enforcement.
  • Headless OpenRouter sign-in — Complete /login over SSH by pasting the redirect URL or authorization code when the loopback callback is unavailable. See OpenRouter.
  • Claude Opus 5 on GitHub Copilot — Use Claude Opus 5 through GitHub Copilot with adaptive thinking and a 1M context window. See GitHub Copilot.

破坏性变更1

  • Upgraded bundled TypeBox aliases to 1.3.7, removing deprecated APIs including Type.Base, Type.Awaited, Type.Promise, Type.AsyncIterator, Type.Iterator, Type.Options, and Value.Mutate, while fixing compiled validation of nullable array tool arguments. Extensions using removed APIs must migrate to supported TypeBox APIs. See Package Dependencies (#7243 by @petrroll).

新增7

  • Added pi auth print-api-key and pi auth print-bearer-token commands for exporting configured credentials to external clients, including automatic OAuth refresh and configurable minimum token validity (#7168).
  • Exposed the session's resolved model scope as ctx.scopedModels to extensions. See Extension Context (#7191 by @pungggi, #7215).
  • Added inherited per-request fetch injection for supported text and image provider transports.
  • Added the inherited "pending" stop reason for partial streaming messages. See Custom Provider Stream Pattern (#7151 by @lucasmeijer).
  • Added inherited raw provider stop reasons across Google, Anthropic, Amazon Bedrock, Mistral, and OpenAI streams; unmapped terminal reasons now surface as provider errors instead of successful stops (#7272).
  • Added manual redirect URL and authorization-code entry to OpenRouter login for remote and headless environments. See OpenRouter (#7114 by @rgarcia).
  • Added inherited Claude Opus 5 support for GitHub Copilot with adaptive thinking and a 1M context window. See GitHub Copilot (#7158 by @jay-aye-see-kay).

变更1

  • Changed inherited OAuth credential resolution to refresh tokens with less than five minutes of validity remaining instead of waiting until expiration (#7168).

修复16

  • Added a status line when the tool output expansion is toggled (#7180).
  • Fixed file-backed SYSTEM.md and APPEND_SYSTEM.md prompts being omitted from the interactive startup context listing. See System Prompt Files (#7096).
  • Fixed context files loading twice when a linked Git worktree is nested under its main repository. See Context Files (#7221 by @arajkumar).
  • Fixed llama.cpp streamed responses reporting zero token usage and leaving session context accounting empty. See llama.cpp (#7258 by @SteveImmanuel).
  • Fixed session replacement and committed tree navigation during an active response to abort and persist the outgoing turn instead of leaving dangling tool calls. See Sessions (#7022 by @tmustier).
  • Fixed failed Git package installs leaving partial directories that blocked clean retries. See Install and Manage (#7210 by @haoqixu).
  • Fixed the /model selector retaining a stale selection while filtering instead of highlighting the top match (#7211 by @christianbasch).
  • Fixed direct RPC bash commands bypassing extension user_bash handlers. See User Bash Events (#7214).
  • Fixed skills, prompts, and themes losing package source metadata after extensions reload resources. See Resource Events (#6968).
  • Fixed cancellation of concurrently running user bash commands so every active command is aborted (#7103 by @yzhg1983).
  • Fixed duplicate messages appearing when extensions switch sessions during interactive startup (#7110 by @yzhg1983).
  • Fixed inherited Qwen Token Plan reasoning models to send their service-specific thinking controls and supported reasoning-effort levels (#6951, #6998).
  • Fixed inherited Z.AI output limits being sent through an unsupported parameter. See Providers (#7174 by @HyeokjaeLee).
  • Fixed explicitly configured Amazon Bedrock profiles being overridden by ambient AWS access keys. See Amazon Bedrock (#7176 by @christianbasch).
  • Fixed inherited image fallback paths overflowing narrow terminals, shortened home-directory paths, and made absolute paths clickable when terminal hyperlinks are available (#7262).
  • Fixed inherited OpenAI-compatible tool calls losing their function arguments when malformed deltas also contain an empty custom object (#7288 by @sunnyyoung).
0.82.1 2026-07-25 新功能 3新增 3变更 3修复 4 Claude Opus 5

新功能3

  • Claude Opus 5 — Available on Anthropic and Amazon Bedrock with adaptive thinking (including xhigh), inference profiles, and prompt caching. See Providers.
  • Anthropic gateway bearer auth — ANTHROPIC_AUTH_TOKEN authenticates against Anthropic-compatible gateways that require Authorization: Bearer, including compaction and branch summaries. See Environment Variables or Auth File.
  • Faster, more resilient model catalogs — pi.dev catalogs revalidate with If-None-Match so unchanged providers answer with an empty 304, and llama.cpp models stay listed across restarts. See llama.cpp.

新增3

  • Exposed the outputPad setting to custom message renderers. See Extensions (#7045 by @xl0).
  • Added inherited ANTHROPIC_AUTH_TOKEN bearer authentication for Anthropic-compatible gateways. See Providers (#5871).
  • Added inherited Claude Opus 5 support for Anthropic and Amazon Bedrock with adaptive thinking, inference profiles, prompt caching, and preserved AWS validation messages (#7081 by @unexge, #7083 by @davidbrai).

变更3

  • Changed pi.dev model catalog refreshes to revalidate with If-None-Match, so unchanged provider catalogs answer with an empty 304 instead of a full download.
  • Changed inherited Radius OAuth device authorization, token exchange, and refresh requests to use the configured gateway directly.
  • Changed inherited model loading errors to append the underlying cause, so auth failures such as OAuth refresh failed for openai-codex report the provider response instead of a bare wrapper message.

修复4

  • Fixed compaction and branch summaries for providers whose authentication resolves entirely to request headers (#5871)
  • Fixed unavailable scoped models being hidden from /models, allowing them to be removed without editing settings manually (#6949, #7032 by @christianklotz).
  • Fixed startup context file discovery to skip directories that match context file names such as AGENTS.md, which produced EISDIR warnings (#7106 by @mrexodia).
  • Fixed the llama.cpp extension to persist its model catalog, so llama.cpp models stay listed before the first successful refresh. See llama.cpp (#7072 by @davidbrai).
0.82.0 2026-07-24 新功能 3新增 6变更 1修复 17 Constrained tool sampling

新功能3

  • Constrained tool sampling — Tools can prefer or require strict JSON Schema sampling or use OpenAI Lark/regex grammars, with model capability metadata preventing unsupported requests. See Constrained Sampling for Tools.
  • OpenRouter and Kimi Code sign-in — Use /login to authorize OpenRouter or a Kimi Code subscription without manually configuring API keys. See OpenRouter.
  • Session-aware, streaming bash integrations — Bash tools receive current session/model metadata, while direct RPC bash commands stream correlated output. See Bash Tool Session Environment and RPC bash events.

新增6

  • Added inherited Tool.constrainedSampling with strict JSON Schema (prefer/require) and OpenAI Lark/regex grammar variants across OpenAI, Anthropic, Amazon Bedrock, Google Gemini, and Mistral. See Constrained Sampling for Tools.
  • Added inherited supportsGrammarTools and supportsStrictTools compatibility flags, expanded supportsStrictMode coverage, and generated model capability metadata to gate constrained sampling.
  • Added inherited Kimi Code subscription OAuth login for the Kimi For Coding provider, including device authorization and automatic token refresh (#6935 by @zaycruz).
  • Added inherited OpenRouter OAuth PKCE login through /login, minting a user-controlled API key. See OpenRouter (#6927 by @rsaryev).
  • Exposed PI_SESSION_ID, PI_SESSION_FILE, PI_PROVIDER, PI_MODEL, and PI_REASONING_LEVEL to commands run by built-in and factory-created bash tools. See Bash Tool Session Environment.
  • Added streaming bash_execution_update events for direct RPC bash commands, correlated with request IDs. See RPC bash events (#6971 by @ananthakumaran).

变更1

  • Changed inherited generated model catalogs to expose only provider-verified reasoning effort levels from models.dev (#6928 by @davidbrai).

修复17

  • Fixed inherited DNS lookup failures such as getaddrinfo, ENOTFOUND, and EAI_AGAIN to trigger automatic assistant retries (#6946 by @christianklotz).
  • Fixed inherited OpenRouter Anthropic cache breakpoints to advance through tool results and enabled cache control for ~anthropic/*-latest aliases (#6941 by @mteam88).
  • Fixed inherited OpenAI Codex WebSocket sessions to retry once without a missing previous-response continuation after previous_response_not_found errors (#6955 by @davidbrai).
  • Fixed TUI debug and crash logs to respect custom agent directories instead of always writing under ~/.pi/agent (#6958 by @davidbrai).
  • Fixed slow Ctrl+G external-editor startup when the system temporary directory contains many entries (#6903 by @christianklotz).
  • Fixed startup resource display to preserve relative paths for sibling npm extensions loaded by a package (#6964 by @davidbrai).
  • Fixed compaction and branch-summary requests to use fresh routing session IDs with prompt caching disabled where supported (#6618 by @tmustier).
  • Fixed explicit self-updates when PI_SKIP_VERSION_CHECK is set (#6977).
  • Fixed scoped model IDs containing brackets to resolve as literal exact matches before glob matching (#6210).
  • Fixed inherited OpenAI and Anthropic provider retry waits to honor abort signals and configured delay limits (#6980 by @petrroll).
  • Fixed fresh installs from preferring bundled model catalogs over newer remote catalogs because package file mtimes were newer (#7016 by @davidbrai).
  • Fixed inherited editor scroll indicators overflowing narrow terminals (#7015 by @christianklotz).
  • Fixed llama.cpp models to use the loaded context window as their output token limit instead of capping it at 16K (#7034 by @christianklotz).
  • Fixed release source archives to include the generated provider model data used to build standalone binaries.
  • Updated the packaged protobufjs dependency to 7.6.5 to address GHSA-j3f2-48v5-ccww (#7005).
  • Fixed /copy on Wayland to fall back to X11 or OSC 52 when wl-copy fails (#7009 by @rkfshakti).
  • Fixed /model to reload updated models.json configuration when opening the model picker (#6999).
0.81.1 2026-07-21 新功能 2新增 1修复 4 Verifiable release source archives

新功能2

  • Verifiable release source archives — GitHub releases now include deterministic, checksummed source archives with instructions for rebuilding standalone binaries. See Building standalone binaries from release source.
  • Resilient compaction and branch summaries — Transient provider failures now follow the configured retry policy, with retry lifecycle events available to interactive, JSON, RPC, and SDK consumers. See Compaction & Branch Summarization and RPC retry events.

新增1

  • Added deterministic, checksummed source archives to GitHub releases with documented standalone binary rebuild instructions (#6913 by @christianklotz).

修复4

  • Fixed compaction and branch summarization to retry transient provider failures using the configured retry policy, with retry lifecycle events exposed to interactive, JSON, RPC, and SDK consumers (#6901 by @davidbrai).
  • Fixed interactive startup waiting for background model catalog refresh while computing the footer provider count.
  • Restored the default stream fallback for extensions using the pre-0.81 agent-core API (#6915).
  • Fixed inherited Kimi K3 models from Moonshot AI and Moonshot AI China to use the OpenAI thinking format and expose reasoning effort support.
0.81.0 2026-07-21 新功能 4新增 6修复 21 Local llama.cpp model management

新功能4

  • Local llama.cpp model management — Connect to a llama.cpp router, search and download Hugging Face models, and explicitly load or unload models with live progress. See llama.cpp.
  • Full provider extensions — Extensions can register complete pi-ai providers with authentication, model refresh, filtering, and custom streaming. See Register New Provider.
  • Qwen Token Plan providers — Use the built-in international and China subscription providers with regional endpoints and API-key authentication. See API Keys.
  • Expanded usage accounting — Tool, compaction, and branch-summary usage is persisted and included in session totals. See Compaction & Branch Summarization.

新增6

  • Added Qwen Token Plan and Qwen Token Plan China to built-in provider setup, default model resolution, and provider documentation (#6858 by @QuintinShaw).
  • Added the get_available_thinking_levels RPC command and RpcClient.getAvailableThinkingLevels() method (#6865 by @cristinaponcela).
  • Exported message and tool execution lifecycle event types from the package root (#6772 by @davidbrai).
  • Added built-in llama.cpp router support with /login connection setup and /llama Hugging Face model search and downloads, explicit loading, unloading, and live progress. See llama.cpp.
  • Added extension registration for complete pi-ai providers, including native authentication, model refresh, filtering, and streaming behavior.
  • Added usage accounting for tools, compaction, and branch summaries in persisted sessions, footer totals, and session statistics (#6671 by @davidbrai).

修复21

  • Updated the packaged brace-expansion dependency to 5.0.7 (#6896 by @davidbrai).
  • Fixed persisted remote model catalogs from overriding newer bundled catalogs after an upgrade.
  • Fixed inherited stored API-key credentials to apply their provider-scoped env values, including Amazon Bedrock profiles (#6864 by @cristinaponcela).
  • Fixed inherited OpenAI-compatible cross-provider replay to keep tool call IDs unique when multiple calls share a provider call ID (#6854 by @cristinaponcela).
  • Fixed inherited Kimi K3 thinking levels to expose low, high, and max, and normalized the k2p7 alias to kimi-for-coding.
  • Fixed inherited OpenCode Go models routed through the OpenAI Responses API.
  • Fixed inherited pi-ai package metadata to avoid repeated consumer lockfile changes (#6812 by @jmfederico).
  • Fixed inherited terminal shutdown to clear the editor's inverted software cursor before restoring the hardware cursor (#6790 by @dam9000).
  • Fixed inherited ANSI-aware text wrapping to recognize CRLF and CR line endings while preserving styles (#6764 by @xz-dev).
  • Fixed inherited editor paste registry corruption after deleting and undoing paste markers, preventing literal or mismatched paste markers in submitted prompts (#6844).
  • Fixed sessionless OpenAI Codex WebSocket requests to use UUIDv7 request IDs (#6834 by @xl0).
  • Fixed inherited GPT-5.6 Codex models to default to the 272K context window, avoiding automatic long-context pricing (#6853 by @aadishv).
  • Fixed messages queued during compaction to preserve steering and follow-up delivery behavior (#6730 by @dannote).
  • Fixed read tool errors being syntax-highlighted as if they were file contents (#6731 by @dannote).
  • Fixed llama.cpp router download progress updates and removed redundant wording from model action confirmations.
  • Moved automatic model catalog network refresh out of startup initialization and into the running interactive and RPC modes.
  • Fixed persisted sessions being read and parsed twice when opened, reducing startup latency for large sessions (#6793).
  • Fixed prompt-template defaults for all arguments (${@:-default} and ${ARGUMENTS:-default}) (#6695).
  • Fixed obsolete custom UI, custom tool, and custom editor examples in the extension documentation (#6735).
  • Fixed Kimi Coding sessions to show API-equivalent implied costs with the subscription indicator.
  • Fixed OpenAI Responses early stream endings to trigger automatic retry instead of ending the agent run (#6727).
0.80.10 2026-07-16 新功能 1修复 4 Kimi Coding thinking compatibility

新功能1

  • Kimi Coding thinking compatibility — Kimi Coding models now use adaptive thinking correctly; K3 exposes its supported max level and supports replaying empty-signature thinking blocks. See Kimi For Coding setup and Model Options.

修复4

  • Fixed inherited Kimi Coding requests to use Anthropic adaptive thinking effort without token budgets, and enabled empty thinking signatures for K3 and kimi-for-coding.
  • Fixed inherited Kimi K3 pricing metadata for Moonshot AI and Moonshot AI China.
  • Fixed inherited Kimi Coding K3 thinking-level metadata to expose only the supported max level (#6737).
  • Fixed inherited catalog generation restoring xAI models removed in 0.80.9 (#6736).
0.80.9 2026-07-16 新功能 1新增 2变更 1修复 2移除 1 Kimi K3 and deferred tool loading

新功能1

  • Kimi K3 and deferred tool loading — Use Kimi K3 across built-in providers, including progressive extension tool activation through Kimi’s native protocol. See Dynamic Tool Loading, OpenAI Compatibility, and the C0 example.

新增2

  • Added inherited Kimi K3 support for Kimi Coding, Moonshot AI, Moonshot AI China, OpenRouter, and Vercel AI Gateway.
  • Added Kimi deferred tool loading for extension-driven tool activation. See Dynamic Tool Loading, OpenAI Compatibility, and the C0 example.

变更1

  • Changed xAI login to use a prefilled device-authorization link labeled “Sign in with SuperGrok or X Premium,” and changed the default xAI model to Grok 4.5 (#6734 by @Jaaneek).

修复2

  • Fixed inherited Kimi K3 output limits for Vercel AI Gateway and OpenRouter models.
  • Fixed cloning or forking a session before its first assistant response to explain that the session must be saved first.

移除1

  • Removed Grok 3, Grok 3 Fast, Grok 4.20 variants, and Grok Code Fast 1 from the built-in xAI model catalog (#6734 by @Jaaneek).
0.80.8 2026-07-16 新功能 3破坏性变更 5新增 6变更 3修复 6 Unified model runtime and provider authentication

新功能3

  • Unified model runtime and provider authentication — ModelRuntime centralizes model configuration, provider-owned /login, and dynamic provider catalogs. See Providers.
  • Live model catalog refresh — /model refreshes configured providers in the background, and pi update --models forces an immediate refresh. See Install and Manage.
  • xAI device-code OAuth and Grok 4.5 Responses support — Sign in to xAI with a device code and use Grok 4.5 with low, medium, or high thinking. See xAI.

破坏性变更5

  • Replaced the SDK's CreateAgentSessionOptions.authStorage and modelRegistry options with the async modelRuntime option. AuthStorage and its storage backends are no longer exported; use ModelRuntime (or a custom pi-ai CredentialStore), or readStoredCredential() for one-off reads of auth.json.
  • Removed redundant ModelRuntime.getAll(), find(), getSnapshot(), and getAuthOptions() projections. Use the pi-ai Models methods getModels(), getModel(), getProviders(), and checkAuth() directly.
  • Replaced SDK request-auth assembly through ModelRegistry.getApiKeyAndHeaders() with ModelRuntime.getAuth(). Passing a provider ID returns provider-scoped auth; passing a model also resolves built-in, models.json, and extension model headers.
  • Changed extension-facing ModelRegistry.refresh() from synchronous void to Promise<void> because models.json loading is asynchronous. Extensions must await it before making synchronous registry reads.
  • Moved canonical dynamic catalog refresh to async ModelRuntime.refresh()/pi-ai Models.refresh(). Legacy extension OAuth modifyModels remains supported as a synchronous compatibility projection after credential initialization.

新增6

  • Added ModelRuntime as the canonical async SDK and internal model/auth facade while preserving the synchronous extension-facing ModelRegistry API. ModelRuntime.create() accepts any pi-ai CredentialStore through its credentials option.
  • Added provider-owned /login discovery directly from registered pi-ai providers, including ambient auth status and informational links.
  • Added file-backed dynamic catalogs in models-store.json, per-provider pi.dev catalog overlays, and Radius gateway support including offline migration from legacy credential-cached catalogs.
  • Added extension provider refreshModels(context) support for dynamic model discovery with optional provider-controlled persistence.
  • Added pi update --models to force an immediate model catalog refresh without updating pi or extensions.
  • Added inherited xAI device-code OAuth login and Grok 4.5 OpenAI Responses support, with low, medium, and high thinking levels (#6651 by @Jaaneek).

变更3

  • Changed ModelRuntime to compose built-in providers, immutable models.json configuration, and extension overlays through ad-hoc pi-ai provider methods.
  • Changed ModelRuntime to own final request assembly: getAuth(model) includes configured model headers, stream methods resolve auth once, and before_provider_headers runs as the Models-only header transform before provider dispatch.
  • Changed /model to render the current model snapshot immediately, refresh configured providers in the background, and update the open selector with partial results or timeout errors.

修复6

  • Fixed configured-provider catalog refresh to parse pi.dev's model-ID keyed responses, throttle checks to once per four hours, send the versioned pi user agent, treat unimplemented routes as unavailable overlays, and show concise refresh status in /model.
  • Fixed adjacent assistant thinking blocks to render as one thinking section.
  • Fixed inherited OpenAI Codex session IDs longer than 64 characters to meet the API limit (#6630).
  • Fixed inherited terminal output to normalize tab characters consistently (#6697 by @xz-dev).
  • Fixed the Windows terminal title after checking npm packages (#6629).
  • Fixed Bun standalone binaries to bundle OAuth adapters for interactive logins.
0.80.7 2026-07-14 破坏性变更 1新功能 3新增 4修复 15 Removed the openai-responses compat.sendSessionIdHeader flag

破坏性变更1

  • Removed the openai-responses compat.sendSessionIdHeader flag from models.json. Session-affinity behavior is now controlled by compat.sessionAffinityFormat ("openai", "openai-nosession", or "openrouter"). Replace sendSessionIdHeader: false with sessionAffinityFormat: "openai-nosession" (#6496 by @petrroll).

新功能3

  • Cache-friendly dynamic tool loading - Extensions can add tools during execution while supported Anthropic and OpenAI Responses models preserve prompt-cache prefixes. See Dynamic Tool Loading.
  • Message copy shortcut - Ctrl+X copies the last assistant message in the transcript or the selected message in /tree, making older and branched messages directly copyable. See Display and Message Queue.
  • Fable 5 xhigh and max thinking - Native xhigh and max thinking levels are available across generated provider catalogs. See Model Options.

新增4

  • Added cache-friendly dynamic tool loading for extension tools activated by tool results. Supported Anthropic and OpenAI Responses models load definitions where they become available, preserving the cached prompt prefix. See Dynamic Tool Loading (#6474).
  • Added inherited native xhigh and max thinking levels for Claude Fable 5 across all generated provider catalogs (#6490 by @davidbrai).
  • Added Ctrl+X to copy the last assistant message, or the selected message in /tree.
  • Added inherited toolChoice support for OpenAI and Codex Responses, including required and named tool selection (#6588 by @xl0).

修复15

  • Fixed inherited OpenRouter model context windows to use the top provider's actual context length (#6481 by @davidbrai).
  • Fixed inherited OpenRouter OpenAI-compatible session IDs to use the x-session-id header instead of OpenAI-specific session-affinity fields (#6496 by @petrroll).
  • Fixed Ctrl+V to paste clipboard text when the pasteboard does not contain an image.
  • Fixed /login amazon-bedrock to prompt for and save a Bedrock API key instead of only displaying ambient AWS credential setup instructions.
  • Fixed inherited Amazon Bedrock ambient AWS credentials to keep using SigV4 authentication, including for custom model IDs (#6532 by @ribelo).
  • Fixed inherited Cloudflare Workers AI and AI Gateway authentication to use ambient account and gateway IDs when stored credentials contain only an API key (#6292 by @markphelps).
  • Fixed inherited legacy terminal decoding for Alt+symbol key combinations such as Alt+, and Alt+. (#6523 by @ribelo).
  • Fixed the GitHub Copilot mai-code-1-flash-picker model to route through the /responses endpoint (#6544 by @petrroll).
  • Fixed branch summaries to work with providers that use ambient authentication instead of API keys (#6595 by @davidbrai).
  • Fixed inherited Amazon Bedrock errors to report unhandled provider stop reasons instead of only An unknown error occurred (#6598 by @davidbrai).
  • Fixed npm package removal when installed packages have conflicting peer dependencies (#6604 by @davidbrai).
  • Fixed inherited Azure OpenAI Responses reasoning replay when encrypted_content appears only in the terminal response event (#6608 by @davidbrai).
  • Fixed inherited Anthropic-compatible proxies that omit usage from message_delta events (#6611 by @davidbrai).
  • Fixed inherited OpenCode OpenAI Responses models to omit the unsupported session-id header while preserving other cache-affinity data (#6645 by @davidbrai).
  • Fixed system prompt cache invalidation across dates by removing the current date from the default prompt (#6621).
0.80.6 2026-07-09 新功能 2新增 3修复 4 max thinking level

新功能2

  • max thinking level - New opt-in thinking level above xhigh, natively supported on GPT-5.6 and adaptive Claude models, available across CLI (--thinking max), SDK, RPC, and model selection. Custom themes can define thinkingMax. See CLI Reference.
  • Input-based pricing tiers - Request-wide input-token pricing tiers for accurate long-context cost accounting (e.g. GPT-5.4/5.5/5.6 long-context rates), also configurable for custom models in models.json and modelOverrides. See Model Configuration.

新增3

  • Added the opt-in max thinking level across CLI, SDK, RPC, model selection, and themes. Custom themes can define thinkingMax; existing themes fall back to thinkingXhigh.
  • Added request-wide input-token pricing tiers to custom model costs in models.json, modelOverrides, and extension-registered providers.
  • Added ~ (home directory) expansion for the shellPath setting (#6470 by @aaronkyriesenbach).

修复4

  • Fixed inherited post-compaction output-token budgeting to ignore stale assistant usage from before the compaction boundary (#6464).
  • Fixed inherited GPT-5.4 and GPT-5.5 long-context cost accounting while retaining the intentional 272K default context limit for models that require an explicit override.
  • Fixed inherited GPT-5.6 metadata to keep direct OpenAI requests in the 272K short-context tier while exposing the Codex backend's 372K context window with long-context pricing, and removed the nonexistent bare gpt-5.6 alias.
  • Fixed inherited Anthropic message conversion to preserve thinking blocks with empty thinking text but a valid signature instead of dropping them, avoiding thinking-block errors on newer Claude models (#6457 by @davidbrai).
0.80.5 2026-07-09 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.80.4 2026-07-09 新功能 4新增 13修复 33移除 1 Prompt cache miss visibility

新功能4

  • Prompt cache miss visibility - Significant cache misses can be shown in transcripts via showCacheMissNotices. See Model & Thinking.
  • Project-local resource configuration - pi config -l and Tab switching manage global vs project-local package resources. See Enable and Disable Resources.
  • Extension lifecycle and provider hooks - Extensions get agent_settled, before_provider_headers, entry renderers, and InlineExtension. See agent_start / agent_end / agent_settled, before_provider_headers, and InlineExtension.
  • New inherited model and transport support - GPT-5.6 metadata, Copilot Claude Sonnet 5, and zstd Codex SSE transport are available through inherited provider support. See Providers and Model Options.

新增13

  • Added inherited OpenAI GPT-5.6 model metadata for gpt-5.6, gpt-5.6-sol, gpt-5.6-terra, and gpt-5.6-luna, plus verified openai-codex support for gpt-5.6-sol, gpt-5.6-terra, and gpt-5.6-luna.
  • Added inherited Claude Sonnet 5 to the GitHub Copilot model catalog (#6200).
  • Added inherited zstd request-body compression for the OpenAI Codex Responses SSE transport.
  • Added /login <provider> support with provider autocomplete.
  • Added public SDK exports for CLI-equivalent model and scoped-model resolution (#6201).
  • Added extension and RPC agent_settled events plus session-level idle waiting for fully settled agent runs (#6363).
  • Added before_provider_headers extension hook support for injecting provider request headers (#6350 by @pmateusz).
  • Added an InlineExtension type for named inline extension factories (#6267 by @any-victor).
  • Added extension entry renderers for persisted display-only session entries that are rendered in interactive mode without being sent to the model context.
  • Added project-local resource override management to pi config, including project mode startup with pi config -l and Tab switching between global and project scopes (#6309).
  • Added inherited InMemorySessionStorage and JsonlSessionStorage exports from the agent harness (#6435).
  • Added inherited custom metadata support in JSONL session headers (#6417 by @ArcadiaLin).
  • Added a showCacheMissNotices setting and /settings toggle for significant prompt-cache miss transcript notices.

修复33

  • Fixed inherited retry classification for gRPC ResourceExhausted provider errors such as NVIDIA NIM transient exhaustion responses (#6449 by @davidbrai).
  • Fixed inherited retry classification for Cloudflare 524 timeout responses (#6239).
  • Fixed inherited GitHub Copilot device-code login polling to wait before the first token poll and to honor server-provided slow_down intervals, avoiding incorrect failures or apparent hangs after browser authorization (#6187).
  • Fixed inherited OpenAI Codex WebSocket sessions to rotate cached connections before the backend's 60-minute limit, avoiding connection-limit failures on long sessions (#6268).
  • Fixed inherited DS4 server context overflow detection for Prompt has ... tokens, but the configured context size is ... tokens errors (#6262).
  • Fixed inherited Fireworks GLM 5.2 Fast to use the OpenAI-compatible endpoint and thinkingLevelMap, aligning it with GLM 5.2 (#6195).
  • Fixed the fork menu to ignore duplicate selection of the same entry (#6430 by @davidbrai).
  • Fixed native clipboard support in Bun standalone releases (#6418 by @davidbrai).
  • Fixed inherited GitHub Copilot extended context window models to use contextWindow: 1000000, preventing premature compaction and under-budgeting (#6439).
  • Fixed inherited editor paste marker accounting when paste markers are deleted or terminal state is cleared (#6397 by @affanali2k3).
  • Fixed null message content from imported transcripts or custom clients to normalize at ingestion boundaries instead of failing during context construction (#6343).
  • Fixed inherited tool calls from length-truncated assistant messages to fail instead of waiting for missing tool results (#6285).
  • Fixed inherited OpenAI Completions and Responses providers to send (no tool output) instead of (see attached image) when a tool result has empty text and no image content.
  • Fixed inherited OpenAI Responses and Azure OpenAI Responses requests to avoid sending max_output_tokens values below the provider minimum (#6265).
  • Fixed inherited Amazon Bedrock prompt-cache points for Claude Fable 5 and Claude Sonnet 5 (#6235).
  • Fixed inherited Amazon Bedrock Claude 5 prompt-cache pricing metadata by removing stale fallback overrides.
  • Fixed inherited OpenAI Codex user-agent construction to synchronously load Node OS metadata, avoiding a startup race that could report pi (browser) in Node/Bun.
  • Fixed pi update for pnpm installations to suggest pruning pnpm's self-update cache when the executable points at a removed cached version (#6279 by @rajp152k).
  • Fixed Xiaomi Token Plan model metadata to follow the upstream models.dev token-plan catalogs, removing unsupported mimo-v2-omni variants (#6204).
  • Fixed startup model selection to skip unauthenticated saved defaults so configured local custom models can be selected instead (#6231).
  • Fixed the question extension example to run question tool calls sequentially so multiple questions in one assistant turn remain answerable (#6189).
  • Fixed /login to report auth storage persistence failures instead of claiming credentials were saved when auth.json is locked (#6223).
  • Fixed split-turn compaction to serialize summary requests so single-concurrency local providers do not fail with 429 errors (#5536).
  • Fixed compaction retained-token budgeting to count context-visible custom messages (#6326).
  • Fixed custom session entries appended during assistant streaming to render before the live assistant message, matching persisted session order.
  • Fixed non-positive or oversized bash tool timeouts to fail with a clear validation error instead of being clamped to an immediate timeout (#6181).
  • Fixed the edit tool schema to allow model-invented extra replacement fields instead of rejecting otherwise valid edits (#6278).
  • Fixed new session resets to clear cached label timestamps (#6354).
  • Fixed auto-retry for Bun fetch socket-drop errors reported as socket connection was closed, so transient provider disconnects do not end headless runs without retrying (#6431).
  • Fixed models.json modelOverrides to apply to extension-registered provider models (#6367).
  • Fixed project context file discovery to use stable parent traversal on Windows so startup no longer hangs while loading AGENTS.md or CLAUDE.md (#6369).
  • Fixed --session-id startup to warn when no existing project session has that id and pi creates a new session (#6407).
  • Fixed /reload help text and docs to consistently mention themes and context files (#6395).

移除1

  • Removed default attribution headers from Vercel AI Gateway requests.
2026-0616 个版本
0.80.3 2026-06-30 新功能 6新增 8变更 2修复 19 Anthropic Claude Sonnet 5 support

新功能6

  • Anthropic Claude Sonnet 5 support - Claude Sonnet 5 is available through inherited Anthropic-compatible and Bedrock provider catalogs with adaptive thinking enabled. See Providers and Model Options.
  • Configurable output spacing - outputPad controls horizontal padding for user messages, assistant messages, and thinking blocks. See Settings.
  • External editor configuration - externalEditor lets Ctrl+G use a configured editor before $VISUAL/$EDITOR fallbacks. See Settings and Keybindings.
  • Richer RPC session tree access - RPC clients can inspect session entries and tree snapshots with get_entries and get_tree. See get_entries and get_tree.
  • Extension session metadata updates - Extensions can observe session name changes through session_info_changed. See session_info_changed.
  • Modern Azure Foundry endpoint support - Azure OpenAI Responses provider setup supports current Microsoft Foundry endpoint URLs. See Azure OpenAI.

新增8

  • Added inherited Anthropic Claude Sonnet 5 model support.
  • Added get_entries and get_tree RPC commands for reading session entries and tree snapshots over RPC (#6078 by @geraschenko).
  • Added a package ./rpc-entry export for launching Pi directly in RPC mode.
  • Added session-name change events for extensions (#6175 by @xl0).
  • Added inherited Azure OpenAI Responses support for modern Microsoft Foundry endpoint URLs (#6004 by @gukoff).
  • Added inherited Usage.reasoning token counts for providers that report reasoning/thinking token usage (#6057).
  • Added an externalEditor settings.json override for Ctrl+G external editor commands, with default fallbacks to Notepad on Windows and nano elsewhere (#6122).
  • Added an outputPad setting for user message, assistant message, and thinking horizontal padding (#6168).

变更2

  • Changed the default OpenAI model to gpt-5.5.
  • Changed inherited OpenAI Codex Responses SSE response-header waits to use the configured HTTP timeout instead of the previous fixed 20 second timeout, reducing false timeouts on slow connections (#4945).

修复19

  • Fixed inherited Claude Sonnet 5 metadata to use adaptive thinking payloads for Anthropic-compatible and Bedrock requests.
  • Fixed inherited generated Xiaomi MiMo model pricing to match current pay-as-you-go pricing from models.dev (#6138).
  • Fixed inherited provider HTTP errors to include response bodies instead of opaque SDK messages (#5832 by @stephanmck).
  • Fixed inherited streamSimple() max-token caps so providers that count input and output against one context window do not reject long requests (#5595).
  • Fixed inherited OpenAI Responses streams to preserve reasoning replay state when output items finish out of order (#6009).
  • Fixed inherited Z.AI preserved thinking requests to send thinking.clear_thinking: false when thinking is enabled, allowing replayed reasoning_content to participate in provider caching (#6083).
  • Fixed pre-prompt compaction to stop after compaction instead of continuing immediately (#6074 by @yzhg1983).
  • Fixed resource notifications to stay before messages when resuming sessions (#6048 by @haoqixu).
  • Fixed startup benchmark timing output to print after TUI shutdown, preserve extension timings, and drain terminal-query replies before stopping benchmark mode (#6030 by @xl0, #6063 by @xl0).
  • Fixed extension tool changes to apply before the next provider request in the same agent run without dropping before_agent_start system-prompt overrides (#6162).
  • Fixed a crash when undici emits an internal client error while terminating a mid-stream HTTP response (#6133).
  • Fixed the compaction event regression test to cover status indicator cleanup and keep CI passing.
  • Fixed interactive status indicators so ending work, retry, compaction, or branch-summary indicators no longer shrink the TUI when clear-on-shrink is enabled (#6026).
  • Fixed --session and SessionManager.open() to reject non-empty invalid session files without overwriting them (#6002).
  • Fixed user-message transcript rendering to keep visible backslashes in Markdown escape sequences such as \" (#6105).
  • Fixed assistant messages stopped by output length to show a visible incomplete-response error (#4290).
  • Fixed --no-session --session-id so ephemeral CLI runs can use deterministic session IDs for provider cache affinity (#6070).
  • Fixed disk BMP image files to be detected, converted to PNG, and attached through read and CLI @file inputs (#6047).
  • Fixed auto-retry for provider stream errors that explicitly tell callers to retry the request (#6019).
0.80.2 2026-06-23 变更 2修复 4 Changed inherited pi-ai ApiKeyCredential to use the auth.jso

变更2

  • Changed inherited pi-ai ApiKeyCredential to use the auth.json-compatible discriminator type: "api_key" and provider-scoped env values instead of type: "api-key" and metadata.
  • Renamed the inherited agent-core public harness shell execution options type from ExecutionEnvExecOptions to ShellExecOptions.

修复4

  • Fixed inherited Anthropic-compatible custom models to use explicit compatibility metadata instead of provider-name heuristics for session-affinity headers and unsupported tool-field omissions.
  • Fixed inherited request-scoped apiKey and env values to participate in provider auth resolution, so providers such as Cloudflare can derive request-specific base URLs from explicit call options (#6021).
  • Restored inherited temporary legacy per-API stream aliases such as streamSimpleOpenAICompletions on the pi-ai compat entrypoint (#6016, #6017).
  • Restored inherited runtime detectCompat fallback in openai-completions for models without explicit compat metadata (#6020).
0.80.1 2026-06-23 修复 3 Fixed inherited Amazon Bedrock scoped AWS_PROFILE endpoint r

修复3

  • Fixed inherited Amazon Bedrock scoped AWS_PROFILE endpoint resolution for built-in inference profile endpoints.
  • Fixed inherited Fireworks Anthropic-compatible requests to apply session-affinity and unsupported tool-field defaults for custom Fireworks models.
  • Fixed inherited Together MiniMax M2.7 metadata to avoid unsupported Together reasoning toggles.
0.80.0 2026-06-23 变更 3修复 11移除 1 Added Ctrl+J as a default newline keybinding alongside Shift

变更3

  • Added Ctrl+J as a default newline keybinding alongside Shift+Enter.
  • Renamed the displayed zai provider label to ZAI Coding Plan (Global) for clarity (#5965).
  • pi-ai's old global API (stream/complete/completeSimple, getModel/getModels/getProviders, registerApiProvider, getEnvApiKey, ...) moved off the @earendil-works/pi-ai root entrypoint to @earendil-works/pi-ai/compat. Extensions are not affected at runtime: the extension loader resolves the pi-ai root to the compat entrypoint (a strict superset), so existing extensions keep working unchanged. Extension sources that typecheck against pi-ai's published types should switch those imports to @earendil-works/pi-ai/compat (or migrate to the new createModels()/provider-factory API). The compat entrypoint and the loader alias will be removed in a future release with a migration guide.

修复11

  • Fixed session names to normalize newline characters before storing or displaying labels (#5999 by @haoqixu).
  • Fixed the session selector to order threaded session trees by the latest activity anywhere in each subtree (#5784 by @Perlence).
  • Fixed extension-related crash and startup-failure reporting to suggest restarting with pi -ne.
  • Fixed inherited OpenAI Responses streams to fail before missing terminal events and fixed context usage and compaction estimates to ignore malformed all-zero assistant usage after truncated responses (#5526 by @dmmulroy).
  • Fixed inherited OpenAI Codex Responses WebSocket sessions to reconnect once when OpenAI's connection limit is reached before output starts (#5973).
  • Fixed inherited Amazon Bedrock endpoint resolution to honor scoped AWS_PROFILE values.
  • Fixed inherited Cloudflare providers to require account/gateway configuration and route built-in compat calls through provider auth.
  • Fixed provider-scoped auth environment values to reach inherited Models/ImagesModels API calls and compat API-key injection.
  • Fixed inherited OpenCode Go GLM-5.2 metadata to expose xhigh reasoning and send the provider's max reasoning effort (#5967).
  • Fixed pi --resume to load user package themes and resolve automatic light/dark theme settings.
  • Fixed models.json custom providers so stored credentials can satisfy auth without a redundant provider-level apiKey (#5953).

移除1

  • Removed inherited selective-provider @earendil-works/pi-ai/base and @earendil-works/pi-agent-core/base entrypoints; use the root packages with explicit Models provider factories instead.
0.79.10 2026-06-22 新功能 2新增 1修复 8 Extension compaction event context

新功能2

  • Extension compaction event context - Extension session_before_compact and session_compact events now include reason and willRetry, so extensions can distinguish manual /compact, threshold auto-compaction, and overflow retry flows. See session_before_compact / session_compact and Custom Summarization via Extensions.
  • Safer update flow - pi update installs the exact checked Pi version, and update notices show the changelog URL, making upgrades more predictable. See Install and Manage.

新增1

  • Added reason and willRetry metadata to extension session_before_compact and session_compact events so extensions can distinguish manual, threshold, and overflow compaction flows (#5962 by @PizzaMarinara).

修复8

  • Fixed the find tool to respect nested git repository boundaries when parent .gitignore rules ignore the nested repo (#5960).
  • Fixed the usage docs slash command table to include /trust and /import (#5959).
  • Fixed inherited OpenAI-compatible streaming to preserve encrypted reasoning_details that arrive before matching tool call deltas (#5114).
  • Fixed broken TUI documentation links to the plan-mode extension example (#5957).
  • Fixed transient extension UI and session-start messages emitted during session replacement or reload so they remain visible, and kept reload input blocked until reload completes (#5943).
  • Fixed the plan-mode example to preserve active custom tools, skip the action prompt when no plan is found, and queue refinement/execution follow-ups correctly from agent_end (#5940).
  • Fixed pi update to install the exact version returned by the Pi update check, make --force reinstall that checked version, fail instead of falling back to an unversioned reinstall when no version is available, and report both the old and updated versions.
  • Fixed update notifications to display the actual changelog URL as the hyperlink text.
0.79.9 2026-06-20 新功能 2新增 1修复 9 Chat-template thinking compatibility

新功能2

  • Chat-template thinking compatibility - OpenAI-compatible custom providers can map Pi thinking levels into chat_template_kwargs, enabling vLLM/Hugging Face chat-template models such as DeepSeek to use provider-native thinking controls. See Custom Provider API Types and OpenAI Compatibility.
  • GLM-5.2 provider improvements - GLM-5.2 now has corrected Fireworks OpenAI-compatible routing and OpenRouter xhigh thinking support, improving /model behavior and high-effort reasoning for GLM-5.2 users. See Model Options.

新增1

  • Added inherited configurable chat-template thinking support for OpenAI-compatible providers that use chat_template_kwargs, such as DeepSeek models behind vLLM (#5673).

修复9

  • Fixed inherited Fireworks GLM-5.2 metadata to use the OpenAI-compatible Chat Completions endpoint with reasoning_effort support (#5923).
  • Fixed same-directory session switches to reuse imported extension modules while preserving fresh extension instances and lifecycle events (#5905).
  • Fixed deep session branches taking quadratic time to build context or branch paths (#5909).
  • Fixed inherited OpenRouter GLM-5.2 metadata to expose xhigh reasoning and send OpenRouter's native xhigh effort (#5770).
  • Fixed inherited Markdown streaming code fence rendering so partial closing fences no longer make code blocks shrink or flicker while content streams (#5846 by @xl0).
  • Fixed fuzzy edit matches to preserve untouched line blocks instead of rewriting the whole file through normalized content (#5899).
  • Fixed bash commands through legacy WSL bash.exe to pass scripts over stdin so shell variables expand in the target bash (#5893).
  • Fixed /model to hide GitHub Copilot models that are unavailable to the authenticated account (#5897).
  • Fixed /model selector search to rank exact provider-prefixed matches before proxy-provider model ID matches (#5892).
0.79.8 2026-06-19 新功能 4新增 4修复 3 Selective provider base entry points

新功能4

  • Selective provider base entry points - SDK users can pair @earendil-works/pi-ai/base and @earendil-works/pi-agent-core/base with explicit provider registration to keep bundled applications from including unused provider transports. See C2 Base Entry Point and C3 Base Entry Point.
  • Mistral prompt caching - Mistral sessions now use provider-side prompt caching with session affinity and cached-token usage/cost accounting. See API Keys and Environment Variables.
  • Post-compaction token estimates - Compact results and compaction events now include estimated post-compaction token counts so clients can show the approximate context reduction. See RPC compact and compaction events.
  • OpenRouter Fusion alias - openrouter/fusion is available as a built-in OpenRouter model alias. See API Keys.

新增4

  • Added inherited @earendil-works/pi-ai/base and @earendil-works/pi-agent-core/base entry points for selective provider registration in bundled applications (#5348 by @FredKSchott).
  • Added inherited Mistral prompt caching using the pi session ID as prompt_cache_key, including cached-token usage and cost accounting (#5854).
  • Added estimated post-compaction token counts to compact results and compaction events (#5877).
  • Added the inherited OpenRouter Fusion alias as openrouter/fusion (#5866 by @dannote).

修复3

  • Updated vulnerable runtime dependencies, including undici and the packaged protobufjs transitive dependency.
  • Fixed compaction to refuse sessions with no eligible messages instead of producing empty summaries (#4811).
  • Fixed successful overflow-triggered auto-compaction to avoid retrying completed assistant responses (#5720).
0.79.7 2026-06-18 新功能 4新增 4变更 3修复 3 Automatic theme mode

新功能4

  • Automatic theme mode - /settings can choose separate light and dark themes and follow terminal color-scheme changes. See Selecting a Theme.
  • Self-only updates by default - pi update now updates pi only, with pi update --all for updating pi and packages together. See Install and Manage.
  • Extension API helpers - extensions can use CONFIG_DIR_NAME for project config paths and import edit diff helpers for edit-style diffs. See C1 and SDK Exports.
  • Warp inline images - Warp terminals now get inline image rendering through Kitty graphics detection. See Image.

新增4

  • Added automatic theme mode so /settings can use separate light and dark themes and follow terminal color-scheme changes (#5874).
  • Added inherited Warp terminal image capability detection so inline images render through Warp's Kitty graphics support (#5841 by @dodiego).
  • Exported CONFIG_DIR_NAME from the coding-agent public API so extensions can resolve project config paths without hardcoding .pi (#5869 by @xl0).
  • Exported edit diff helpers (generateDiffString, generateUnifiedPatch, and EditDiffResult) from the public API for extensions that need edit-style diffs (#5756 by @xl0).

变更3

  • Changed bare pi update to update only pi, added pi update --all for updating pi and extensions together, and clarified extension update prompts.
  • Reserved / in theme names for automatic light/dark theme settings.
  • Updated extension docs, examples, runtime help, trust prompts, and config labels to use the configured project config directory instead of hardcoded .pi paths.

修复3

  • Fixed RPC unknown-command errors to include the request id so clients do not hang waiting for a response (#5868).
  • Fixed /model autocomplete and model selection searches to match provider/model queries regardless of whether the provider or model token is typed first.
  • Fixed the tree navigator to horizontally pan deep entries so the selected item remains readable (#5830).
0.79.6 2026-06-16 修复 2 Fixed HTTP dispatcher configuration to preserve a caller's d

修复2

  • Fixed HTTP dispatcher configuration to preserve a caller's deliberate fetch override instead of reinstalling the undici global fetch over it.
  • Fixed inherited OpenCode Go DeepSeek V4 thinking-off requests to send the provider's thinking: { type: "disabled" } compatibility parameter.
0.79.5 2026-06-16 新功能 3新增 4变更 1修复 9 Provider-scoped API key environments

新功能3

  • Provider-scoped API key environments - auth.json API key entries can now include env overrides for provider-specific Cloudflare, Azure OpenAI, Google Vertex, Amazon Bedrock, cache retention, and proxy settings without changing the project shell. See Auth File.
  • Global HTTP proxy setting - Configure httpProxy once in global settings to apply HTTP_PROXY and HTTPS_PROXY to Pi-managed HTTP clients. See Network.
  • Vercel AI Gateway attribution - Vercel AI Gateway requests now include Pi attribution headers by default. See API Keys.

新增4

  • Added Vercel AI Gateway request attribution headers (http-referer and x-title) for Vercel AI Gateway models (#5798 by @rwachtler).
  • Added an xp footer marker when experimental features are enabled.
  • Added a global httpProxy setting that applies as HTTP_PROXY and HTTPS_PROXY for Pi-managed HTTP clients (#5790).
  • Added auth.json API key env values so provider-specific environment overrides can be scoped to Pi and propagated to inherited provider configuration (#5728).

变更1

  • Updated the vendored Markdown parser used by HTML session exports to marked 18.0.5.

修复9

  • Fixed inherited OpenAI Responses streaming to tolerate null message content from OpenAI-compatible servers before tool calls (#5819).
  • Fixed inherited OpenCode DeepSeek V4 thinking requests to avoid sending both thinking and reasoning_effort (#5818).
  • Fixed device-code login to stop opening the browser automatically.
  • Fixed inherited editor Cursor Up handling so non-empty drafts jump to the start of the line before browsing input history (#5789 by @4h9fbZ).
  • Fixed inherited Z.AI GLM-5.2 thinking requests to send reasoning_effort with the provider's high/max effort mapping (#5770).
  • Fixed successful pi update on Windows to exit naturally instead of calling process.exit(0), avoiding a Node.js/libuv assertion after version-check network requests (#5805).
  • Fixed inherited Google and google-vertex Gemini model metadata to map latest aliases to the current models, add Gemini 3.5 Flash for Vertex, correct Gemini 2.5 Flash Vertex cache pricing, and remove shut-down Vertex preview models (#5761).
  • Fixed the session selector to stay open and show the all-sessions empty state when both current-folder and all-scope session lists are empty (#5747).
  • Fixed inherited Moonshot AI China model metadata to include Kimi K2.7 Code, and omitted unsupported thinking-off payloads for Kimi K2.7 Code models (#5760).
0.79.4 2026-06-15 新功能 2新增 2修复 14 Automatic first-run theme selection

新功能2

  • Automatic first-run theme selection - pi detects the terminal background on first run and defaults to the dark or light theme. See Selecting a Theme.
  • Standalone binary integrity checksums - GitHub release assets now include SHA256SUMS files for verifying standalone binary downloads. See Quickstart Install.

新增2

  • Added SHA256SUMS integrity files to standalone binary GitHub release assets (#5739).
  • Added first-run interactive theme detection from the terminal background (#5385 by @vegarsti).

修复14

  • Fixed bash tool output collection to keep draining stdout/stderr after the child exits while descendants still write, avoiding truncated late output (#5753 by @Mearman).
  • Fixed /tree help rendering to show compact wrapped controls instead of truncating them on narrow terminals (#5055).
  • Fixed SIGTERM/SIGHUP interactive shutdown to keep signal handlers installed until terminal cleanup completes, preventing signal-exit from re-sending the signal and leaving the terminal in raw/Kitty keyboard mode (#5724).
  • Fixed extensions documentation to clarify that pi.getActiveTools() returns active tool names while pi.getAllTools() returns tool metadata (#5729).
  • Fixed question and questionnaire extension examples to wrap long prompt, option, and help text instead of truncating it (#5708 by @xl0).
  • Fixed package commands such as pi list, pi install, and pi update to terminate after completing even if an extension leaves background handles open (#5687).
  • Fixed pi update for pnpm global installs whose configured global-bin-dir no longer matches the active pnpm home (#5689).
  • Fixed npm package specs that use ranges or tags (for example @^1.2.7) so installed package resources still load instead of being treated as mismatched exact pins (#5695).
  • Fixed inherited Anthropic 1-hour prompt-cache write cost accounting to price 1-hour cache writes at 2x input instead of the 5-minute cache-write rate (#5738 by @theBucky).
  • Fixed inherited GitHub Copilot Claude adaptive-thinking effort metadata to match manually checked Copilot model capabilities (#4637).
  • Fixed inherited OpenCode/OpenCode Go completion model metadata to omit long-retention cache fields for routes that reject prompt_cache_retention (#5702).
  • Fixed inherited overlay compositing over CJK wide characters so borders stay aligned when an overlay starts inside a full-width cell (#5297).
  • Fixed inherited WezTerm inline Kitty image rendering during full redraw fallbacks so image padding rows are reserved before the placement is drawn without regressing tall-image placement (#5618, #4415).
  • Fixed custom provider config so plain uppercase API key and header values remain literals instead of being treated as legacy environment references; use explicit $ENV_VAR syntax for environment variables (#5661).
0.79.3 2026-06-13 修复 1 Fixed inherited OpenAI GPT-5.4/GPT-5.5 and OpenAI Codex GPT-

修复1

  • Fixed inherited OpenAI GPT-5.4/GPT-5.5 and OpenAI Codex GPT-5.4/GPT-5.4 mini/GPT-5.5 context window metadata to use the observed 272k-token Codex backend limit, avoiding a billing hazard from prompts above Codex's accepted limit (reported by @trethore).
0.79.2 2026-06-12 新功能 1新增 2修复 17 Clearer Bedrock validation guidance

新功能1

  • Clearer Bedrock validation guidance - Amazon Bedrock data retention validation errors now link to AWS data retention documentation. See Amazon Bedrock.

新增2

  • Added an experimental first-time setup flow behind PI_EXPERIMENTAL=1 that asks for a dark/light theme choice (preselecting the detected appearance) and opt-in analytics data sharing on first launch with the default agent directory; opting in stores a trackingId in settings.json (#5587 by @vegarsti).
  • Added AWS data retention documentation links to inherited Amazon Bedrock unsupported data retention mode validation errors (#5561 by @unexge).

修复17

  • Fixed project trust detection to ignore global ~/.pi/agent state when running from $HOME, and made pi update use only saved or explicit project trust without prompting (#5619).
  • Fixed experimental first-time setup to skip forked sessions instead of rerunning the setup prompts (#5627 by @vegarsti).
  • Fixed inherited OpenAI-compatible context overflow detection for parenthesized maximum context length (N) errors (#5677).
  • Fixed inherited OpenAI GPT-5.4/GPT-5.5 and OpenAI Codex GPT-5.4/GPT-5.4 mini/GPT-5.5 context window metadata to match current OpenAI limits (#5644).
  • Fixed inherited Anthropic refusal stops to preserve provider stop_details explanations in error messages (#5666 by @rwachtler).
  • Increased the inherited OpenAI Codex Responses SSE response-header timeout to 20 seconds to reduce false-positive stalls while retaining the bounded wait introduced for zero-event hangs (#4945).
  • Fixed inherited Claude Fable 5 thinking-off requests to omit Anthropic's unsupported thinking.type: "disabled" payload (#5567 by @tmustier).
  • Fixed inherited late tool progress callbacks after tool settlement to be ignored instead of emitting stale tool_execution_update events (#5573).
  • Fixed inherited user-message transcript rendering so standalone + messages no longer render as - (#5657).
  • Fixed inherited slash-separated fuzzy queries so provider/model completions remain matchable after insertion.
  • Fixed inherited WezTerm inline Kitty image rendering so reserved row clears do not erase all but the top strip of tool image previews (#5618).
  • Fixed inherited editor wrapping for CJK text to break at character boundaries instead of leaving large trailing gaps (#5585 by @haoqixu).
  • Fixed inherited loose Markdown list rendering to preserve blank-line separation between list items (#5562 by @Perlence).
  • Fixed --model resolution for authenticated custom model IDs whose slash prefix matches an unauthenticated built-in provider (#5643).
  • Fixed /fork to keep session parent chains connected when the forked path contains labels (#5669).
  • Fixed /share and /export HTML exports to use the active fallback theme when the configured custom theme no longer exists (#5596).
  • Fixed custom fallback model IDs with :<thinking> suffixes to preserve the requested thinking level when the provider template model does not advertise reasoning (#5560 by @haoqixu).
0.79.1 2026-06-09 新功能 4新增 6修复 17 Claude Fable 5

新功能4

  • Claude Fable 5 - Claude Fable 5 is now available on the Anthropic and Amazon Bedrock providers, with adaptive thinking and xhigh effort support.
  • Prompt template defaults - Prompt templates can use default positional arguments such as ${1:-7} for optional values. See Prompt Template Arguments.
  • Configurable project trust defaults - defaultProjectTrust lets users choose whether unresolved project trust asks, always trusts, or never trusts by default, and extensions can inspect effective trust decisions. See Project Trust and C1.
  • Natural extension autocomplete triggers - Extension autocomplete providers can declare trigger characters such as # or $ so suggestions open without slash-command prefixes. See Autocomplete Providers.

新增6

  • Added default-value expansion for prompt template positional arguments, e.g. ${1:-7} (#5553 by @dannote).
  • Added areExperimentalFeaturesEnabled feature guard to allow users to opt in to early features (#5547 by @vegarsti).
  • Added ctx.isProjectTrusted() for extensions to observe the effective project trust decision, including temporary trust decisions (#5523).
  • Added a global defaultProjectTrust setting to choose whether unresolved project trust asks, always trusts, or never trusts by default.
  • Added extension autocomplete trigger character support for ctx.ui.addAutocompleteProvider() wrappers (#4703).
  • Added Claude Fable 5 model support inherited from @earendil-works/pi-ai for the Anthropic and Amazon Bedrock providers, with adaptive thinking and xhigh effort support.

修复17

  • Fixed inherited Amazon Bedrock inference profile ARN region resolution to prefer the ARN's embedded region over AWS_REGION (#5527 by @AJM10565).
  • Fixed inherited IME hardware cursor positioning while slash-command autocomplete is visible (#5283 by @smoosex).
  • Fixed inherited z.ai thinking-off requests to send the provider's thinking: { type: "disabled" } compatibility parameter (#5330).
  • Fixed inherited OpenCode completions model metadata to send explicit maxTokens as max_tokens (#5331).
  • Fixed inherited Moonshot Kimi thinking-off requests to send the provider's thinking: { type: "disabled" } compatibility parameter (#5531).
  • Fixed inherited Azure OpenAI Responses requests to disable server-side response storage (#5530).
  • Fixed inherited Azure GPT-5.4 and GPT-5.5 context window metadata to 1,050,000 tokens, matching Azure Foundry deployments instead of OpenAI's 272k limit (#5559).
  • Fixed inherited OpenAI and Azure GPT-5 Pro maxTokens metadata to 128,000, correcting an upstream value that duplicated the input sub-limit as the output limit (#5559).
  • Fixed inherited prompt history navigation to restore the current draft when returning from history browsing (#5494).
  • Fixed inherited wrapping for mixed Latin and CJK text so unspaced CJK runs can break at grapheme boundaries without leaving large trailing gaps (#5495).
  • Fixed extension OAuth login prompts to keep previous submitted prompt rows stable instead of mirroring the active input value (#5433).
  • Fixed /reload to apply updated steeringMode and followUpMode settings to the current session (#5377).
  • Fixed invalid models.json syntax to skip startup config migrations and report the normal file-path-aware models error instead of a raw JSON parse stack trace (#5418).
  • Fixed GitHub release notes and interactive changelog links to resolve package-relative documentation URLs correctly (#5516).
  • Fixed CLI help and version output, including plain redirected --help/--version output and simplified list/config help text.
  • Fixed /new from ephemeral sessions to keep the new session ephemeral instead of persisting it by default (#5045).
  • Clarified custom model docs that name and modelOverrides.name do not replace model IDs in the footer or primary model lists (#4841).
0.79.0 2026-06-08 新功能 4新增 5修复 13 Project trust for local inputs

新功能4

  • Project trust for local inputs - Pi now asks before loading project-local settings, resources, instructions, and packages, with saved decisions and --approve / --no-approve controls for non-interactive modes. See Project Trust.
  • Extension-controlled trust decisions - Global and CLI extensions can handle project_trust, decide, remember, or defer project trust before project-local resources load. See C1.
  • Cache-hit visibility in the footer - The interactive footer now shows the latest prompt cache hit rate (CH). See Interactive Mode.
  • Richer SDK and RPC extension surfaces - Public exports now include RPC extension UI request/response types and package asset path helpers. See Extension UI Protocol and SDK Exports.

新增5

  • Added a project_trust extension event so global and CLI extensions can decide or defer project trust during startup and runtime cwd switches.
  • Added project trust gating for project-local settings, resources, instructions, and packages (#5332).
  • Added the latest prompt cache hit rate to the interactive footer.
  • Exported RPC extension UI request and response types from the public API (#5455).
  • Exported coding-agent package asset path helpers from the public API (#5415).

修复13

  • Fixed package exports by removing the stale ./hooks subpath that pointed at non-existent build output.
  • Fixed inherited TUI rendering to clear stale lines when content shrinks to zero.
  • Fixed inherited autocomplete suggestions to refresh after editor cursor movement (#5499 by @Roman-Galeev).
  • Fixed /reload to persist project trust when an implicitly trusted session creates a project .pi directory.
  • Fixed project trust input discovery to traverse parent directories portably.
  • Fixed inherited intermittent Shift+Enter handling by making Kitty keyboard protocol fallback response-driven instead of timeout-driven (#5188).
  • Fixed the compaction summarization system prompt to use neutral AI assistant wording for non-coding agents (#5401).
  • Fixed models.json schema support and inherited OpenAI Responses custom-provider handling for compat.supportsDeveloperRole: false (#5456).
  • Fixed inherited prompt history navigation to place the cursor at the start when browsing upward and at the end when browsing downward (#5454).
  • Fixed tmux setup documentation to require tmux 3.5 for extended-keys-format csi-u and document the tmux 3.2-3.4 fallback (#5432).
  • Fixed inherited OpenRouter routing preferences on OpenAI-compatible custom providers to work when the custom provider base URL does not point directly at OpenRouter (#5347).
  • Fixed built-in tool expand hints to style closing parentheses consistently (#5359).
  • Fixed skill-wrapped prompts to insert spacing between skill instructions and the user message (#5371 by @Perlence).
0.78.1 2026-06-04 新功能 2新增 6修复 16 More built-in provider coverage

新功能2

  • More built-in provider coverage - Added Ant Ling and NVIDIA NIM provider setup, plus MiniMax-M3 support for the direct MiniMax providers. See Providers.
  • Richer extension context - Extensions can use ctx.mode and ctx.getSystemPromptOptions() to adapt behavior across TUI, RPC, JSON, and print modes and inspect base system prompt inputs. See Extensions.

新增6

  • Added containerization documentation and a Gondolin extension example for routing built-in tools into a local micro-VM.
  • Added Ant Ling provider selection and setup documentation.
  • Added MiniMax-M3 model support inherited from @earendil-works/pi-ai for the minimax and minimax-cn direct providers (#5313).
  • Added NVIDIA NIM provider selection, setup documentation, and direct NIM request attribution headers.
  • Added ctx.mode to extension contexts so extensions can distinguish TUI, RPC, JSON, and print mode.
  • Added ctx.getSystemPromptOptions() for extension commands to inspect the current base system prompt inputs (#5306 by @xl0).

修复16

  • Fixed temporary extension package installs to use a private ~/.pi/agent/tmp/extensions directory with 0700 permissions instead of os.tmpdir()/pi-extensions.
  • Fixed git package source handling to reject unsafe host/path components and keep managed clone paths inside install roots.
  • Fixed stored XSS in HTML session exports by sanitizing Markdown link and image URLs with a scheme allow-list after stripping control characters.
  • Fixed SDK embedding in bundled Node apps failing with ENOENT when package.json is not present next to the bundle entrypoint. The package metadata reader now gracefully handles missing package.json by using defaults, enabling createAgentSession() without requiring package-adjacent files at runtime (#5226).
  • Fixed HTTP timeout setting not being respected for non-Codex providers (e.g., llama.cpp via OpenAI-compatible API). The httpIdleTimeoutMs setting (set via /settings HTTP timeout) now applies as the default SDK request timeout for all providers that support it, not just OpenAI Codex Responses. Disabling the timeout (HTTP timeout = false) now correctly disables SDK timeouts for all supported providers by sending a maximum int32 value (effectively infinite) instead of 0, since SDKs treat timeout=0 as an immediate timeout (#5294).
  • Fixed inherited Amazon Bedrock requests to replace blank required user/tool-result text with a placeholder and skip blank replay text blocks (#4975).
  • Fixed inherited Anthropic Claude Opus 4.7+ requests to suppress deprecated temperature parameters (#5251 by @yzhg1983).
  • Fixed inherited OpenAI GPT-5.5 generated metadata to omit unsupported minimal thinking (#5243).
  • Fixed inherited OpenRouter Kimi K2.6 thinking replay and developer-role instruction handling (#5309).
  • Fixed inherited OpenRouter reasoning instruction requests to preserve the system role when required (#5221 by @PriNova).
  • Fixed inherited overlay focus restoration so non-capturing overlays remain interactive after UI rerenders and explicit focus release (#5235 by @nicobailon).
  • Fixed inherited tab width accounting in column slicing and overlay compositing so tab-containing output cannot exceed the terminal width (#5218).
  • Fixed opening and listing very large JSONL session files by reading session entries line-by-line instead of materializing the full file as one string (#5231).
  • Fixed the footer branch display in WSL /mnt/... repositories to refresh after branch changes (#5264 by @psoukie).
  • Fixed renderShell: "self" tool renderers that emit no component lines leaving a blank chat row (#5299).
  • Restored inherited NVIDIA Qwen 3.5 122B NIM model support.
2026-0516 个版本
0.78.0 2026-05-29 新功能 2新增 6修复 10 Named startup sessions

新功能2

  • Named startup sessions - --name / -n sets the session display name before startup across interactive, print, JSON, and RPC modes. See Naming Sessions and Session Options.
  • Clickable file tool paths - built-in file tool titles render OSC 8 file:// hyperlinks when the terminal supports them, including supported tmux clients.

新增6

  • Exported convertToPng for extension authors (#5167 by @xl0).
  • Exported parseArgs and type Args for extension authors (#5202 by @xl0).
  • Added --name / -n to set the session display name at startup (#5153).
  • Added a resume command hint when exiting interactive sessions (#5176 by @yzhg1983).
  • Added OSC 8 file:// hyperlinks to file paths shown in built-in file tool titles (#5189 by @mpazik).
  • Added custom Amazon Bedrock request header support inherited from @earendil-works/pi-ai (#5178 by @stephanmck).

修复10

  • Clarified the WezTerm/WSL IME hardware cursor docs to state that cursor visibility remains opt-in (#5200).
  • Fixed the GitLab Duo custom provider example to use adaptive thinking for Claude models, expose xhigh thinking, and include newer verified model IDs (#5201).
  • Fixed Bun release archive creation to install and copy the matching @mariozechner/clipboard base package and native sidecars (#5184).
  • Fixed early interactive input typed before the prompt loop starts so it is buffered instead of dropped (#5195 by @yzhg1983).
  • Fixed OpenRouter Moonshot Kimi K2.6 requests to use system instead of unsupported developer messages (#5159).
  • Fixed OpenCode Go Kimi K2.6 thinking requests to send thinking objects instead of invalid string values, and fixed OpenCode Zen Grok Build thinking requests to omit unsupported reasoning_effort (#5169).
  • Fixed OpenAI Codex Responses SSE streams to abort response body reads after terminal events.
  • Fixed OpenCode Kimi K2.6 generated metadata to use Anthropic-style thinking metadata instead of invalid reasoning-effort parameters.
  • Fixed OSC 8 hyperlinks to pass through tmux when the client supports them (#5189 by @mpazik).
  • Fixed ANSI text wrapping to avoid stack overflows on very long wrapped lines (#5185).
0.77.0 2026-05-28 新功能 4新增 4修复 18 Claude Opus 4.8 support

新功能4

  • Claude Opus 4.8 support - Adds Anthropic Claude Opus 4.8 metadata and updates Opus adaptive-thinking coverage.
  • Selective tool disablement - --exclude-tools / -xt disables specific built-in, extension, or custom tools while leaving the rest available. See Tool Options.
  • Headless Codex subscription login - /login can use device-code auth for ChatGPT Plus/Pro Codex subscriptions. See Subscriptions and OpenAI Codex.
  • Streaming-aware extension input - extensions can distinguish idle prompts, mid-stream steers, and queued follow-ups with InputEvent.streamingBehavior. See Input Events.

新增4

  • Added --exclude-tools / -xt to disable specific built-in, extension, or custom tools while leaving the rest available (#5109).
  • Added OpenAI Codex subscription device-code login as a selectable headless alternative while keeping browser login as the default (#4911 by @vegarsti).
  • Added streamingBehavior to extension input events so extensions can distinguish idle prompts from mid-stream steers and queued follow-ups (#5107 by @DanielThomas).
  • Added Claude Opus 4.8 model metadata for Anthropic and updated Opus adaptive-thinking coverage to use it.

修复18

  • Fixed startup timing output so readPipedStdin no longer includes createAgentSessionRuntime work (#4829).
  • Fixed OpenRouter DeepSeek V4 xhigh reasoning metadata to preserve OpenRouter's native effort instead of sending DeepSeek's max effort (#4801).
  • Fixed custom session directories so current-folder resume/continue lookups stay scoped to the active cwd while all-session listings cover the custom directory.
  • Fixed SIGTERM/SIGHUP exits to run extension session_shutdown cleanup and restore the terminal: signal-triggered shutdown now emits session_shutdown before any terminal writes, and SIGHUP no longer hard-exits, so extension resources (e.g. sockets) are released even when the terminal is gone (#5080).
  • Fixed keyboard protocol negotiation to ignore mismatched or delayed terminal responses, avoiding false Kitty keyboard protocol detection (#5091 by @mitsuhiko).
  • Fixed Windows startup crashes under MSYS2 ucrt64 Node.js by updating the native clipboard addon to napi-rs 3.x (#5028).
  • Fixed API key and header config resolution to treat plain strings as literals, support $ENV_VAR / ${ENV_VAR} interpolation and $! bang escaping, and require explicit env syntax for config files, avoiding Windows case-insensitive env matches corrupting literal keys (#5095).
  • Fixed session disposal to abort in-flight agent, compaction, branch summary, retry, and bash work (#5029 by @TerminallyChilI).
  • Fixed pi.getAllTools() to expose each tool's promptGuidelines for extensions that need per-tool guideline attribution (#4879).
  • Fixed OpenAI Codex Responses replay after switching from Anthropic extended-thinking sessions by generating unique fallback message item IDs for converted thinking/text blocks (#5148).
  • Fixed Anthropic-compatible replay for providers that return empty thinking signatures by adding an opt-in allowEmptySignature compatibility flag (#4464).
  • Fixed OpenAI and OpenRouter GPT-5.5 Pro thinking level metadata to expose only supported medium, high, and xhigh efforts.
  • Fixed OpenCode Go Kimi K2.6 thinking-off requests to send thinking: "none" (#5078).
  • Fixed Xiaomi Token Plan model metadata to omit unsupported mimo-v2-flash variants (#5075).
  • Fixed follow-up messages queued by agent_end extension handlers to drain before the agent becomes idle (#5115 by @DanielThomas).
  • Fixed extension input events to report streamingBehavior only for prompts actually queued during streaming (#5107 by @DanielThomas).
  • Fixed system prompt tool-selection guidance to avoid preferring unavailable file exploration tools (#5132).
  • Fixed fenced diff code blocks and other highlight.js scopes to keep theme-aware syntax colors after the cli-highlight replacement (#5092).
0.76.0 2026-05-27 新功能 4新增 2修复 16 Explicit session IDs for automation

新功能4

  • Explicit session IDs for automation - --session-id <id> lets scripts create or resume an exact project-local session. See Sessions.
  • RPC bash output can stay out of model context - RPC clients can pass excludeFromContext to bash for commands whose output should not be sent with the next prompt. See RPC mode.
  • More predictable provider retries and timeouts - Codex WebSocket/SSE waits are bounded, and retry.provider.maxRetries controls provider retries instead of hidden SDK defaults. See Retry settings.
  • Better terminal editing across environments - Apple Terminal Shift+Enter, Windows/JetBrains capability detection, and Unicode-aware word navigation improve interactive editing. See Terminal setup and Keybindings.

新增2

  • Added --session-id to let CLI callers use an exact project-local session ID, creating it if missing (#4874).
  • Added excludeFromContext flag to the bash RPC command for parity with the internal executeBash API (#5039).

修复16

  • Fixed user message transcript rendering to preserve user-authored ordered-list markers (#5013).
  • Fixed self-update commands to bypass npm, pnpm, and Bun minimum release age gates for explicit pi update runs (#4929).
  • Fixed context token estimates to count user image attachments consistently with tool result images (#4983).
  • Fixed httpIdleTimeoutMs to apply to OpenAI Codex Responses WebSocket idle waits, added websocketConnectTimeoutMs for bounded WebSocket connect waits, and added a 10s Codex SSE response-header timeout (#4945).
  • Fixed RpcClient to reject pending requests and consume stdin pipe errors when the child process exits unexpectedly (#4764).
  • Fixed managed npm extension updates to avoid package managers installing or resolving pi host packages as peer dependencies (#4907).
  • Fixed RPC mode raw stdout writes to retry transient backpressure errors and flush queued protocol output during shutdown (#4897).
  • Fixed OpenAI Codex Responses cache-affinity headers to send session-id instead of proxy-incompatible session_id (#4967).
  • Fixed openai-codex/gpt-5.3-codex-spark model metadata to use its 128k context window (#4969).
  • Fixed OpenRouter/Poolside context overflow detection for maximum allowed input length errors (#4943).
  • Fixed provider retry controls so retry.provider.maxRetries is honored, SDK retries default to 0, and quota/billing 429s are not retried behind Pi's retry handling (#4991 by @mitsuhiko).
  • Fixed Apple Terminal Shift+Enter by detecting local macOS modifier state when Terminal.app sends plain Return.
  • Fixed Windows Terminal capability detection to enable OSC 8 hyperlinks, preserving clickable long URLs across wrapped lines (#4923).
  • Fixed JetBrains terminal capability detection to enable truecolor while disabling unsupported OSC 8 hyperlinks (#5037 by @Perlence).
  • Fixed editor and input word navigation/deletion to use Unicode word boundaries while preserving ASCII punctuation boundaries (#5022 by @haoqixu, #5067 by @haoqixu, #5068 by @haoqixu).
  • Fixed the development docs AGENTS.md link to point at the pi-mono guidelines (#5041).
0.75.5 2026-05-23 新功能 4新增 3变更 3修复 16 Cleaner read tool output

新功能4

  • Cleaner read tool output - Collapsed read tool cards now show only the read line by default, while Ctrl+O still expands the full file content.
  • Faster file tools on Windows - Built-in file tools now use async filesystem operations during streaming, and image resizes run off the main TUI thread in a worker.
  • More reliable package updates - pi update and git package installs now reconcile pinned git refs and keep package settings intact. See Packages.
  • Custom Anthropic-compatible adaptive thinking - Custom provider model configs can opt into adaptive-thinking Claude behavior with compat.forceAdaptiveThinking. See Custom providers and Models.

新增3

  • Added compat.forceAdaptiveThinking support to custom Anthropic-compatible model configuration docs and validation (#4797 by @mbazso).
  • Added a standard unified patch to edit tool result details for SDK consumers (#4821).
  • Added a Codex subscription login method selector with device-code auth for headless environments.

变更3

  • Changed collapsed read tool cards to show only the read line until expanded (#4916).
  • Replaced the inherited optional koffi dependency for Windows VT input with a tiny vendored native helper, reducing install size while preserving Shift+Tab handling (#4480).
  • Changed the root development install documentation to use npm install --ignore-scripts (#4868).

修复16

  • Fixed pi update to reconcile git-pinned packages to their configured ref (#4869).
  • Fixed package/resource path handling for Windows and glob/pattern resolution (#4873 by @mitsuhiko).
  • Fixed config pattern matching to resolve patterns from the correct base directory (#4898 by @haoqixu).
  • Fixed theme pickers to list themes by their content name instead of file stem (#4830 by @Perlence).
  • Fixed OpenCode Zen/Go requests to send per-session OpenCode routing headers (#4847).
  • Fixed Amazon Bedrock provider loading under strict package managers by inheriting the declared @smithy/node-http-handler dependency from @earendil-works/pi-ai (#4842).
  • Fixed inherited Amazon Bedrock Claude requests to send the model output token cap by default, avoiding Bedrock's 4096-token default truncation (#4848).
  • Fixed exported session HTML to escape quote characters in attribute values (#4832).
  • Fixed GitHub Copilot device-code login to keep opening the verification URL in browser-capable environments while ignoring browser launch failures for headless use (#4788 by @vegarsti).
  • Fixed git package installs to reconcile existing checkouts to the requested ref and update package settings without losing filters (#4870).
  • Published a 0.74.2 rescue release that tells Node 20 users to upgrade Node before updating to newer Pi versions (#4876).
  • Fixed final bash tool cards to avoid rendering duplicate full-output truncation paths (#4819).
  • Fixed bash tool truncation line counts to ignore the trailing newline as an extra output line (#4818).
  • Fixed footer home-directory abbreviation to avoid shortening sibling paths that only share the same prefix (#4878).
  • Fixed macOS Bun release binaries to resolve the native clipboard sidecar so Ctrl+V image paste can load @mariozechner/clipboard (#4307).
  • Fixed coding-agent tools to avoid synchronous filesystem operations during streaming and moved image resizing off the main TUI thread (#4756 by @mitsuhiko).
0.75.4 2026-05-20 新功能 1新增 2变更 5修复 16 Hardened npm install and release path

新功能1

  • Hardened npm install and release path - Pi now ships the CLI with a generated shrinkwrap for transitive dependencies, blocks accidental lockfile changes, verifies dependency pinning and lifecycle-script allowlists in checks, disables lifecycle scripts for self-update and local release installs where supported, and smoke-tests isolated npm and Bun installs before release. See Supply-chain hardening.

新增2

  • Added interactive update notes after pi update runs, so users can see the installed version's changelog before continuing (#4724 by @mitsuhiko).
  • Exported image resize utilities from the package root for SDK consumers (#4775 by @xl0).

变更5

  • Changed source syntax to avoid TypeScript constructs that require JavaScript emit, keeping core sources compatible with Node.js strip-only TypeScript checks.
  • Removed web UI workspace references from the CLI package and dropped the package-level development watch script.
  • Published npm installs now include an npm-shrinkwrap.json to lock transitive dependencies for the CLI package.
  • Improved terminal theme detection for light/dark and truecolor handling.
  • Changed self-update package-manager commands to disable lifecycle scripts during reinstall.

修复16

  • Fixed the system prompt to tell models to resolve pi docs and examples under the absolute package paths before reading topic-specific relative references (#4752).
  • Fixed extension ctx.abort() during tool-call preflight to stop later confirmations and restore queued interactive input like Escape (#4276).
  • Fixed AgentSession retry, compaction, and event settlement to use the awaited agent lifecycle instead of a separate event queue, and added willRetry to agent_end session events.
  • Fixed forked session runtime state to keep the active session id aligned with the fork target (#4799 by @Perlence).
  • Fixed the subagent extension's parallel mode to return useful per-task output and failed-task diagnostics to the parent model instead of 100-character previews (#4710).
  • Fixed Windows local bash execution to hide helper console windows when launched from background SDK processes (#4699).
  • Fixed managed npm extension folders to set cloud-sync ignore metadata where supported (#4763).
  • Fixed HTTP idle timeout configuration so long-running provider streams can avoid premature idle disconnects (#4759 by @mitsuhiko).
  • Fixed default system prompt boundaries to use explicit XML tags for clearer file separation (#4709 by @herrnel).
  • Fixed HTML share/export sidebar clicks for shared tool entries to scroll to the rendered tool call (#4664 by @yzhg1983).
  • Fixed theme palettes to set explicit text colors and avoid terminal-default color drift.
  • Fixed truecolor detection to align terminal image rendering and interactive theme decisions.
  • Fixed loader indicator startup inherited from @earendil-works/pi-tui so initialization cannot run before frames are available.
  • Fixed OpenAI-compatible default output token requests inherited from @earendil-works/pi-ai to avoid reserving impossible context windows on servers such as vLLM (#4675).
  • Fixed OpenAI prompt cache keys inherited from @earendil-works/pi-ai to stay within the 64-character provider limit (#4720).
  • Fixed Windows npm-family package commands for fnm-managed Node.js installs that expose both extensionless Unix scripts and .cmd shims (#4793).
0.75.3 2026-05-18 修复 1 Fixed undici 8 HTTP/2 destroyed-session races crashing the N

修复1

  • Fixed undici 8 HTTP/2 destroyed-session races crashing the Node CLI by preserving the previous HTTP/1.1-only fetch dispatcher behavior (#4681).
0.75.2 2026-05-18 修复 7 Fixed Bun-compiled release binaries failing to start when Bu

修复7

  • Fixed Bun-compiled release binaries failing to start when Bun's built-in undici shim lacks npm undici's install export (#4661 by @dmasiero).
  • Fixed Xiaomi MiMo generated model metadata to replay assistant tool-call messages with reasoning_content for thinking-mode multi-turn requests, inherited from @earendil-works/pi-ai (#4678).
  • Fixed Windows external editor handoff so vim/nvim can receive input after opening from the TUI (#4612).
  • Fixed Windows npm self-updates to move loaded native dependency packages out of the active install before reinstalling pi (#4157).
  • Fixed pi update --self detection for pnpm v11 global installs whose package path resolves through the pnpm store (#4647).
  • Fixed Windows pnpm self-updates to resolve pnpm command shims and run through pnpm instead of requiring manual updates (#4157).
  • Fixed Windows npm-family command execution to use cross-spawn instead of parsing .cmd shim internals (#4665).
0.75.1 2026-05-18 修复 8移除 1 Fixed config selectors to scale their visible row count to t

修复8

  • Fixed config selectors to scale their visible row count to terminal height (#4243 by @samjonester).
  • Fixed Anthropic-compatible API-key requests to ignore unrelated ANTHROPIC_AUTH_TOKEN environment values, avoiding invalid bearer credentials for providers such as Xiaomi MiMo inherited from @earendil-works/pi-ai (#4342).
  • Fixed Amazon Bedrock message conversion to skip unknown content blocks instead of failing the stream, inherited from @earendil-works/pi-ai (#4223).
  • Fixed Azure OpenAI Responses and OpenAI Responses error formatting to prefix HTTP status codes onto errorMessage, so transient 5xx and 429 errors are correctly matched by the agent-level auto-retry classifier inherited from @earendil-works/pi-ai (#4232).
  • Fixed OpenCode Go Kimi reasoning replay by normalizing streamed reasoning fields back to reasoning_content for OpenCode Go only, inherited from @earendil-works/pi-ai (#4251).
  • Fixed Xiaomi MiMo model metadata to use the OpenAI-compatible endpoints and openai-completions API, restoring multi-turn thinking/tool-call sessions inherited from @earendil-works/pi-ai (#4505).
  • Fixed JSON parse failures for compressed fetch responses under Node 26.0 by installing undici fetch globals alongside pi's global dispatcher (#4650, #4652, #4653).
  • Fixed npm-family package commands on Windows to avoid shell argument splitting when install prefixes contain spaces (#4623).

移除1

  • Removed non-working OpenAI Codex fast model variants inherited from @earendil-works/pi-ai.
0.75.0 2026-05-17 破坏性变更 1修复 7 Raised the minimum supported Node.js version to 22.19.0.

破坏性变更1

  • Raised the minimum supported Node.js version to 22.19.0.

修复7

  • Fixed compaction summary calls to use custom agent stream functions, preserving proxy-backed LLM routing (#4484).
  • Fixed system prompt and context file boundaries to use explicit XML tags instead of Markdown headings, reducing inconsistent boundary ingestion by models (#4541 by @herrnel).
  • Fixed OpenAI Codex generated model metadata to use the current upstream model list inherited from @earendil-works/pi-ai (#4603 by @mattiacerutti).
  • Fixed GitHub Copilot GPT model thinking metadata inherited from @earendil-works/pi-ai to map unsupported minimal thinking to low (#4622 by @mattiacerutti).
  • Fixed user-scoped npm pi packages to install under ~/.pi/agent/npm/ instead of npm's global package root, avoiding permission errors with system-managed Node installs (#4587).
  • Fixed Mistral requests failing after the global fetch proxy/timeout workaround by removing the custom fetch override and using undici 8 dispatcher support instead (#4619).
  • Fixed default output token requests for models whose advertised output limit is effectively their full context window, avoiding impossible provider requests inherited from @earendil-works/pi-ai (#4614).
0.74.1 2026-05-16 新功能 4新增 3修复 23 Image generation support

新功能4

  • Image generation support - Added image generation APIs, generated image model metadata, and built-in OpenRouter image generation support inherited from @earendil-works/pi-ai.
  • Together AI provider - Added Together AI as a built-in provider with /login API-key auth, default model resolution, and setup docs. See README.md#providers--models and docs/providers.md.
  • Windows ARM64 standalone binaries - Added standalone release artifacts for Windows ARM64.
  • Improved terminal and markdown rendering - Added markdown list indentation, task-list checkbox rendering, large markdown robustness, and inline image placement fixes inherited from @earendil-works/pi-tui.

新增3

  • Added image generation support from @earendil-works/pi-ai, including image generation APIs, image model metadata, and built-in OpenRouter image generation support (#3887 by @cristinaponcela).
  • Added Together AI to built-in provider setup, /login API-key auth, and default model resolution (#3624 by @Nutlope).
  • Added Windows ARM64 standalone binary release artifacts (#4458 by @brianmichel).

修复23

  • Fixed Node 26 OpenAI-compatible streams timing out after five idle minutes by routing global fetch through pi's undici dispatcher (#4519).
  • Fixed pnpm global package installs by resolving the global package root from pnpm's layout.
  • Fixed macOS clipboard access errors under sandboxed pasteboard denial so they do not abort the process (#4492).
  • Fixed the scoped model startup hint to show the configured model-cycle keybinding (#4508).
  • Fixed resource path display to disambiguate package/resource names that collide across package locations.
  • Fixed fd auto-download on macOS x86_64 by pinning the last release that ships an Intel macOS binary (#4559).
  • Fixed skill diagnostics to stop warning when a skill name differs from its parent directory (#4534).
  • Fixed prompt template argument parsing to split unquoted multiline input on newlines (#4553).
  • Fixed --resume session listing to cap in-flight session metadata loads and avoid OOM on large session histories (#4583).
  • Fixed interactive error messages to render with trailing spacing so reload errors do not run into resource listings (#4510).
  • Fixed .agents package provenance metadata to survive package-manager scans.
  • Fixed nested code fences in the Termux setup documentation so the example AGENTS.md renders correctly (#4503).
  • Fixed tool output expansion while extension confirmation dialogs are focused (#4429).
  • Fixed auto-retry for Anthropic streams that end before message_stop (#4433).
  • Fixed compaction summary calls to clamp requested output tokens to model limits.
  • Fixed uncaught interactive-mode exceptions to restore the terminal before exiting (#4426 by @ofa1).
  • Fixed ANSI stripping to match strip-ansi behavior after dependency removal.
  • Fixed UUIDv7 sequence generation shared by session IDs after dependency removal.
  • Fixed OpenRouter cached-token usage accounting, Fireworks caching compatibility, and OpenAI Codex WebSocket proxy handling inherited from @earendil-works/pi-ai.
  • Fixed markdown list wrapping, task-list checkboxes, large markdown rendering, WezTerm Kitty keyboard escape handling, and short-viewport inline image placement inherited from @earendil-works/pi-tui.
  • Fixed theme sharing across package scopes so extensions do not crash with Theme not initialized (#4333).
  • Fixed keybinding hints to show Option instead of Alt on macOS (#4289).
  • Fixed the interactive update notification to render the changelog as an OSC 8 hyperlink when the terminal supports hyperlinks (#4280).
0.74.0 2026-05-07 变更 1 Updated repository links and package references for the move

变更1

  • Updated repository links and package references for the move to earendil-works/pi-mono and @earendil-works/* package scopes.
0.73.1 2026-05-07 新功能 3新增 1变更 3修复 11 Self-update support for the npm scope migration

新功能3

  • Self-update support for the npm scope migration: pi update --self now supports the upcoming package rename from @mariozechner/pi-coding-agent to @earendil-works/pi-coding-agent. After the new package is published, existing global installs can update through the normal self-update flow; pi will uninstall the old global package and install the package name returned by the version check endpoint.
  • Interactive OAuth login selection: OAuth providers can now present multiple login choices in /login, enabling provider-specific interactive authentication flows. See Providers.
  • JSONC-style models.json parsing: models.json now allows comments and trailing commas, making custom provider and model configuration easier to maintain. See Providers and Custom Providers.

新增1

  • Added interactive login selection support so OAuth providers can present multiple login choices (#4190 by @mitsuhiko).

变更3

  • Changed pi update --self to honor the active package name returned by the Pi version check endpoint, defaulting to the current package when omitted and uninstalling the old global package before installing a renamed package.
  • Changed extension loading to use upstream jiti 2.7 instead of the @mariozechner/jiti fork (#4244 by @pi0).
  • Changed models.json parsing to allow comments and trailing commas (#4162 by @julien-c).

修复11

  • Fixed pi -p treating prompts that start with YAML frontmatter as extension flags instead of user messages (#4163).
  • Fixed pending tool results not updating in the live TUI after toggling thinking block visibility while the tool is running (#4167).
  • Fixed /copy reporting success on Linux without writing the clipboard on Wayland-only compositors (Hyprland, Niri, ...) by skipping the X11-only native addon on Linux and routing through wl-copy/xclip/xsel instead (#4177).
  • Fixed HTML session exports to strip skill wrapper XML from rendered user messages (#4234 by @aliou).
  • Fixed OpenAI-compatible chat completion streams that interleave content and tool-call deltas in the same choice.
  • Fixed OpenAI Codex OAuth refresh failures writing directly to stderr while the TUI is active (#4141).
  • Fixed OpenAI Codex Responses requests to send a non-empty system prompt (#4184).
  • Fixed Kimi For Coding model resolution for the Kimi K2 P6 alias (#4218).
  • Fixed Kitty inline image redraws to stay within TUI-owned terminal regions and avoid writing below the active viewport.
  • Fixed Kitty inline image rendering by letting the terminal allocate image ids and bounding parsed image ids to valid values.
  • Fixed inline image capability detection to disable inline images in cmux terminals.
0.73.0 2026-05-04 新功能 3破坏性变更 1新增 1变更 1修复 8 Xiaomi MiMo API billing and regional Token Plan providers

新功能3

  • Xiaomi MiMo API billing and regional Token Plan providers - xiaomi now uses API billing, with separate xiaomi-token-plan-{cn,ams,sgp} providers. See docs/providers.md#api-keys and README.md#providers--models. (#4112 by @Phoen1xCode)
  • Incremental bash output streaming - Bash tool output now appears while commands run instead of only after completion. (#4145)
  • Compact read rendering - Interactive read output for Pi docs, context files, and skills is collapsed by default and shows selected line ranges.

破坏性变更1

  • Switched the built-in xiaomi provider from Token Plan AMS to Xiaomi's API billing endpoint, and renamed its /login display from "Xiaomi MiMo Token Plan" to "Xiaomi MiMo". XIAOMI_API_KEY now refers to the API billing key from platform.xiaomimimo.com. Users on Token Plan should switch to the appropriate xiaomi-token-plan-* provider and set the corresponding env var (#4112 by @Phoen1xCode).

新增1

  • Added three Xiaomi MiMo Token Plan regional providers visible in /login: xiaomi-token-plan-cn (XIAOMI_TOKEN_PLAN_CN_API_KEY), xiaomi-token-plan-ams (XIAOMI_TOKEN_PLAN_AMS_API_KEY), xiaomi-token-plan-sgp (XIAOMI_TOKEN_PLAN_SGP_API_KEY). Each defaults to mimo-v2.5-pro (#4112 by @Phoen1xCode).

变更1

  • Changed read tool rendering to collapse Pi documentation, AGENTS/CLAUDE context files, and SKILL.md contents by default in interactive output.

修复8

  • Fixed generated OpenAI-compatible model metadata for Qwen 3.5/3.6 and MiniMax M2.7, so those models work through the built-in provider catalog (#4110 by @jsynowiec).
  • Fixed Bedrock Claude Opus 4.7 xhigh thinking requests by preserving the provider's native effort value.
  • Fixed OpenAI Codex WebSocket transport to fall back to SSE when setup fails before streaming starts, and surface transport diagnostics in the assistant message (#4133).
  • Fixed OpenAI Codex WebSocket transport keeping --print and JSON mode processes alive after the response by closing cached WebSocket sessions during session shutdown (#4103).
  • Fixed compact read tool calls to render directly and include selected line ranges in interactive output.
  • Fixed interactive sessions to exit when terminal input is lost instead of continuing in a broken state.
  • Fixed bash tool output to stream incrementally while commands run instead of waiting for command completion (#4145).
  • Fixed selector and autocomplete fuzzy ranking to prioritize exact matches.
0.72.1 2026-05-02 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.72.0 2026-05-01 新功能 5破坏性变更 1新增 3修复 3 Xiaomi MiMo Token Plan provider

新功能5

  • Xiaomi MiMo Token Plan provider - New Anthropic-compatible provider with XIAOMI_API_KEY auth, default model (mimo-v2.5-pro), and /login display. See docs/providers.md. (#4005 by @Phoen1xCode).
  • Model thinking level metadata - Models can now declare which thinking levels they support via thinkingLevelMap, replacing the old reasoningEffortMap. See docs/models.md#thinking-level-map and docs/custom-provider.md. (#3208).
  • Custom provider base URL overrides - pi.registerProvider() now respects per-model baseUrl settings. See docs/custom-provider.md. (#4063).
  • Post-turn stop callback - Agent loop can now exit gracefully after a completed turn via shouldStopAfterTurn. See C1.
  • Self-update detection fix - pi now correctly identifies and applies available updates. (#3942, #3980, #3922).

破坏性变更1

  • Replaced compat.reasoningEffortMap in models.json and pi.registerProvider() model definitions with model-level thinkingLevelMap (#3208). Migration: move old mappings from compat.reasoningEffortMap to thinkingLevelMap. Use string values for provider-specific thinking values and null for unsupported pi levels that should be hidden and skipped by cycling. See docs/models.md#thinking-level-map and docs/custom-provider.md.

新增3

  • Added Xiaomi MiMo Token Plan provider support with XIAOMI_API_KEY, default model resolution, /login display support, and provider documentation (#4005 by @Phoen1xCode).
  • Added model-level thinkingLevelMap support in models.json and pi.registerProvider(), allowing models to expose only the thinking levels they actually support (#3208).
  • Added shouldStopAfterTurn agent loop callback for post-turn stop control, inherited from @mariozechner/pi-agent-core. See C2.

修复3

  • Fixed the default transport setting to use auto, allowing OpenAI Codex to use cached WebSocket context when available (#4083).
  • Fixed pi.registerProvider() to honor per-model baseUrl overrides (#4063).
  • Fixed self-update detection so pi correctly identifies when a newer version is available and applies updates (#3942, #3980, #3922).
0.71.1 2026-05-01 新增 1 Added websocket-cached to the transport setting options for

新增1

  • Added websocket-cached to the transport setting options for the OpenAI Codex provider used with ChatGPT subscription auth. This keeps the same WebSocket open for a session and, after the first request, sends only the new conversation items instead of resending the full chat history when possible.
2026-0425 个版本
0.71.0 2026-04-30 破坏性变更 1新功能 5新增 9修复 22移除 2 Removed built-in Google Gemini CLI and Google Antigravity su

破坏性变更1

  • Removed built-in Google Gemini CLI and Google Antigravity support. Existing configurations using those providers must switch to another supported provider.

新功能5

  • Cloudflare AI Gateway provider support with CLOUDFLARE_API_KEY/CLOUDFLARE_ACCOUNT_ID/CLOUDFLARE_GATEWAY_ID, default model resolution, and /login display. See docs/providers.md#cloudflare-ai-gateway. (#3856 by @mchenco).
  • Moonshot AI provider support with MOONSHOT_API_KEY, default model resolution, and /login display.
  • Mistral Medium 3.5 built-in model support. See docs/providers.md#api-keys. (#4009 by @technocidal).
  • Extension APIs can replace finalized message_end messages, wrap custom editor factories via ctx.ui.getEditorComponent(), and observe thinking level changes. See docs/extensions.md#message_start--message_update--message_end, docs/extensions.md#widgets-status-and-footer, and docs/extensions.md#thinking_level_select.
  • PI_CODING_AGENT_SESSION_DIR configures session storage from the environment. See docs/usage.md#environment-variables.

新增9

  • Added Cloudflare AI Gateway as a built-in provider with CLOUDFLARE_API_KEY/CLOUDFLARE_ACCOUNT_ID/CLOUDFLARE_GATEWAY_ID setup, default model resolution, /login display support, and provider documentation (#3856 by @mchenco).
  • Added Moonshot AI as a built-in provider with MOONSHOT_API_KEY setup, default model resolution, and /login display support.
  • Added Mistral Medium 3.5 built-in model support via @mariozechner/pi-ai (#4009 by @technocidal).
  • Added routed OpenAI-compatible response model metadata in assistant messages, so providers such as OpenRouter can expose the concrete model used (#3968 by @purrgrammer).
  • Added PI_CODING_AGENT_SESSION_DIR as an environment equivalent to --session-dir (#4027).
  • Added message_end extension result support for replacing finalized messages, enabling extensions to override assistant usage cost (#3982).
  • Added top-level name support to pi.registerProvider() so extension-registered providers can show a friendly name in /login (#3956).
  • Added ctx.ui.getEditorComponent() so extensions can wrap the currently configured custom editor factory (#3935).
  • Added a thinking_level_select extension event for observing thinking level changes (#3888).

修复22

  • Fixed WSL clipboard image paste by passing the PowerShell save path directly instead of through a custom environment variable (#2469).
  • Fixed Google Vertex Gemini 3 tool call replay for unsigned tool calls (#4032).
  • Fixed blocked edit tool results rendering the rejection reason twice after interactive extension confirmation (#3830).
  • Fixed extension-triggered thinking level changes refreshing the interactive editor border immediately (#3888).
  • Fixed the coding-agent README See Also link to point at @mariozechner/pi-agent-core (#4023).
  • Fixed grep and find tool argument injection for flag-like search patterns (#4018).
  • Fixed PowerShell shell command output on Windows by only spawning detached processes on Unix (#4013 by @picasso250).
  • Fixed Bun package manager node_modules discovery when npmCommand is configured to use Bun (#3998 by @thirtythreeforty).
  • Fixed edit and edit-preview access failures to report filesystem errors correctly (#3955 by @rwachtler).
  • Fixed ProcessTerminal sizing to use COLUMNS and LINES before falling back to 80x24 (#4004).
  • Updated @anthropic-ai/sdk to clear GHSA-p7fg-763f-g4gf audit findings (#3992).
  • Updated @mariozechner/clipboard to an attested release so package managers with trust policies do not reject installs (#3946).
  • Fixed project context discovery to load AGENTS.MD files in addition to AGENTS.md (#3949).
  • Fixed /handoff to use compacted session context instead of pre-compaction raw messages (#3945).
  • Fixed DeepSeek V4 Flash xhigh thinking support so requests map to DeepSeek's max reasoning effort (#3944).
  • Fixed Anthropic streams that end before message_stop to be treated as errors instead of successful partial responses (#3936).
  • Fixed generated OpenAI-compatible DeepSeek V4 reasoning compatibility outside the direct DeepSeek provider (#3940).
  • Fixed idle follow-up submission to clear the editor like normal message submission (#3926).
  • Fixed editor rendering artifacts for Thai Sara Am and Lao AM vowel characters (#3904).
  • Fixed DeepSeek V4 Flash and V4 Pro pricing metadata to match current official rates (#3910).
  • Updated the sandbox extension example lockfile to resolve the vulnerable lodash-es transitive dependency (#3901).
  • Fixed DeepSeek prompt cache hits to be tracked from OpenAI-compatible usage responses (#3880).

移除2

  • Removed the discontinued Qwen CLI OAuth custom provider extension example (#3832 by @4h9fbZ).
  • Removed Google Gemini CLI and Google Antigravity built-in login, default model, documentation, and example extension support.
0.70.6 2026-04-28 新功能 2新增 1变更 1修复 8 Cloudflare Workers AI provider support with CLOUDFLARE_API_K

新功能2

  • Cloudflare Workers AI provider support with CLOUDFLARE_API_KEY/CLOUDFLARE_ACCOUNT_ID setup. See docs/providers.md#api-keys. (#3851 by @mchenco)
  • Pi update checks now use pi.dev and identify Pi with a pi/<version> user agent. See docs/packages.md. (#3877 by @mitsuhiko)

新增1

  • Added Cloudflare Workers AI as a built-in provider with CLOUDFLARE_API_KEY/CLOUDFLARE_ACCOUNT_ID setup, default model resolution, /login support, and provider documentation (#3851 by @mchenco).

变更1

  • Changed Pi version checks to identify Pi with a pi/<version> user agent (#3877 by @mitsuhiko).

修复8

  • Fixed config selector scroll indicators to show item counts instead of line counts (#3820 by @aliou).
  • Fixed exported HTML to escape embedded image data and session metadata, preventing crafted session content from injecting markup (#3819 by @justinpbarnett, #3883 by @justinpbarnett).
  • Fixed Bun-based package manager startup by locating global node_modules relative to Bun's install layout (#3861 by @thirtythreeforty).
  • Fixed Bedrock inference profile capability checks by normalizing profile ARNs to the underlying model name.
  • Fixed file discovery to fall back to fdfind when fd is unavailable.
  • Fixed pi update to skip self-update reinstalls when the installed version is already current (#3853).
  • Fixed Cloudflare Workers AI attribution headers to honor the install telemetry setting.
  • Fixed pi update --self detection and execution for Windows package-manager shim installs, including symlinked global package roots, and print the manual fallback command when self-update fails (#3857).
0.70.5 2026-04-27 修复 1 Fixed HTML export preserving ANSI-renderer trailing padding

修复1

  • Fixed HTML export preserving ANSI-renderer trailing padding as extra blank wrapped lines.
0.70.4 2026-04-27 修复 1 Fixed packaged pi startup failing because the session select

修复1

  • Fixed packaged pi startup failing because the session selector imported a source-only utility path.
0.70.3 2026-04-27 新功能 4新增 4修复 17 pi update can now update pi itself in addition to installed

新功能4

  • pi update can now update pi itself in addition to installed pi packages. See docs/packages.md. (#3680 by @mitsuhiko)
  • Azure Cognitive Services endpoint support for Azure OpenAI Responses deployments. See docs/providers.md#api-keys. (#3799 by @marcbloech)
  • Suppressible Anthropic extra-usage billing warning via warnings.anthropicExtraUsage in /settings. See docs/settings.md. (#3808)
  • Extension-controlled working row visibility via ctx.ui.setWorkingVisible(), allowing extensions to hide the built-in loader row and render custom working state. See docs/extensions.md and examples/extensions/border-status-editor.ts. (#3674)

新增4

  • Added pi update support for updating pi itself in addition to installed pi packages (#3680 by @mitsuhiko).
  • Added Azure Cognitive Services endpoint support for Azure OpenAI Responses base URLs (#3799 by @marcbloech).
  • Added warnings.anthropicExtraUsage and a /settings warnings submenu to suppress the Anthropic extra usage billing warning (#3808)
  • Added ctx.ui.setWorkingVisible() so extensions can hide the built-in interactive working loader row without reserving layout space, plus a border-status editor example that moves working state into a custom editor border (#3674)

修复17

  • Fixed duplicate printable characters from Kitty keyboard protocol CSI-u plus raw character input on layouts such as Italian (#3780).
  • Fixed API-key environment discovery and Bun startup to fall back to /proc/self/environ when Bun's sandbox leaves process.env empty (#3801 by @mdsjip).
  • Fixed Bun sandboxed package-manager commands when process.env is empty (#3807 by @mdsjip).
  • Fixed symlinked packages, resources, skills, and sessions being duplicated in selectors and loaders (#3818 by @aliou).
  • Fixed Bedrock prompt-caching and adaptive-thinking capability checks for inference profile ARNs (#3527 by @anirudhmarc).
  • Fixed OpenAI Codex Responses default verbosity to low when no verbosity is specified.
  • Stopped sending empty tools arrays to providers that reject them when tools are disabled (#3650 by @HQidea).
  • Fixed Anthropic SSE parsing to ignore unknown proxy events such as OpenAI-style done terminators (#3708).
  • Fixed provider registration with override-only models.json entries to preserve built-in model lists (#3651).
  • Fixed /login to show auth supplied by models.json provider definitions.
  • Fixed HTML export whitespace around extension-rendered tool output and expandable output hints.
  • Fixed bash executor temp output streams leaking file descriptors when output was truncated by line count (#3786)
  • Fixed extension pi.setSessionName() updates to refresh the interactive terminal title immediately (#3686)
  • Fixed /tree cancellation via session_before_tree leaving the session stuck in compaction state (#3688)
  • Fixed Escape interrupt handling when extensions hide the built-in working loader row (#3674)
  • Fixed coding-agent test expectations for current default models and missing-auth guidance.
  • Fixed long local-LLM SSE streams aborting at 5 minutes with UND_ERR_BODY_TIMEOUT by disabling undici bodyTimeout/headersTimeout on the global dispatcher; provider SDKs continue to enforce their own deadlines via retry.provider.timeoutMs (#3715)
0.70.2 2026-04-24 修复 1 Fixed provider retry/timeout forwarding to omit undefined pr

修复1

  • Fixed provider retry/timeout forwarding to omit undefined provider request controls, avoiding downstream SDK validation errors such as timeout must be an integer when retry.provider.timeoutMs is not configured (#3627)
0.70.1 2026-04-24 新功能 2新增 1修复 7 DeepSeek provider support with V4 Flash/Pro models and DEEPS

新功能2

  • DeepSeek provider support with V4 Flash/Pro models and DEEPSEEK_API_KEY authentication. See README.md#providers--models and docs/providers.md#api-keys.
  • Provider request timeout/retry controls via retry.provider.{timeoutMs,maxRetries,maxRetryDelayMs}, useful for long-running local inference and provider SDK retry behavior. See docs/settings.md#retry. (#3627)

新增1

  • Added DeepSeek to built-in provider setup, default model resolution, and provider documentation.

修复7

  • Fixed /copy to avoid unbounded OSC 52 writes and clipboard races that could break terminal rendering or panic the native clipboard addon (#3639)
  • Fixed extension flag docs to show pi.getFlag() using registered flag names without the CLI -- prefix (#3614)
  • Fixed provider retry/timeout settings wiring by adding retry.provider.{timeoutMs,maxRetries,maxRetryDelayMs}, migrating legacy retry.maxDelayMs, and forwarding provider controls into streamSimple request options (#3627)
  • Fixed Windows git package installs to bypass cmd.exe for native git commands, so install paths containing spaces no longer break pi install git:... with fatal: Too many arguments (#3642)
  • Fixed DeepSeek V4 session replay 400 errors by sending DeepSeek-compatible thinking controls and replayed assistant reasoning_content fields (#3636)
  • Fixed GPT-5.5 generated context window metadata to use the observed 272k limit.
  • Fixed CSI-u Ctrl+letter decoding inside bracketed paste, so pasted modified-key escape sequences no longer become literal editor text (#3623 by @Exrun94)
0.70.0 2026-04-23 新功能 4破坏性变更 1新增 3变更 2修复 28 Searchable auth provider login flow: the /login provider sel

新功能4

  • Searchable auth provider login flow: the /login provider selector now supports fuzzy search/filtering, making it faster to find providers when many are configured. See docs/providers.md. (#3572 by @mitsuhiko)
  • GPT-5.5 Codex support: openai-codex/gpt-5.5 is available as a model option, including xhigh reasoning support and corrected priority-tier pricing.
  • Terminal progress indicators are now opt-in: OSC 9;4 progress reporting during streaming/compaction is off by default and can be toggled via terminal.showTerminalProgress in /settings (#3588)
  • --no-builtin-tools / createAgentSession({ noTools: "builtin" }) now correctly disables only built-in tools while keeping extension tools active. See docs/extensions.md and README.md (#3592)

破坏性变更1

  • Disabled OSC 9;4 terminal progress indicators by default. Set terminal.showTerminalProgress to true in /settings to re-enable (#3588)

新增3

  • Added searchable auth provider login flow with fuzzy filtering in the provider selector (#3572 by @mitsuhiko)
  • Added GPT-5.5 Codex model
  • Added auth source labels in /login so provider entries can show when auth comes from --api-key, an environment variable, or custom provider fallback without exposing secrets.

变更2

  • Updated default model selection across providers to current recommended models.
  • Improved stale extension context errors after session replacement or reload to tell extension authors to avoid captured pi/command ctx and use withSession for post-replacement work.

修复28

  • Fixed /model selector cancellation to request render instead of incorrectly triggering login selector.
  • Changed login, OAuth, and extension selectors for more consistent styling.
  • Added Amazon Bedrock setup guidance to /login and updated /model copy to refer to configured providers instead of only API keys.
  • Improved no-model and missing-auth warnings to point users to /login for OAuth or API key setup.
  • Fixed /quit shutdown ordering to stop the TUI before extension UI teardown can repaint, preserving the final rendered frame while still emitting session_shutdown before process exit.
  • Fixed SettingsManager.inMemory() initial settings being lost after reloads triggered by SDK resource loading (#3616)
  • Fixed models.json provider compatibility to accept compat.supportsLongCacheRetention, allowing proxies to opt out of long-retention cache fields when needed while long retention is enabled by default when requested (#3543)
  • Fixed --thinking xhigh for openai-codex gpt-5.5 so it is no longer downgraded to high.
  • Fixed git package installs with custom npmCommand values such as pnpm by avoiding npm-specific production flags in that compatibility path (#3604)
  • Fixed first user messages rendering without spacing after existing notices such as compaction summaries or status messages (#3613)
  • Fixed the handoff extension example to use the replacement-session context after creating a new session, avoiding stale ctx errors when it installs the generated prompt (#3606)
  • Fixed session replacement and /quit teardown ordering to run host-owned extension UI cleanup synchronously after session_shutdown handlers complete but before invalidating the old extension context, preventing stale extension UI from rendering against a disposed session (#3597 by @vegarsti)
  • Fixed crash on /quit when an extension registers a custom footer whose render() accesses ctx, by tearing down extension-provided UI before invalidating the extension runner during shutdown (#3595)
  • Fixed auto-retry to treat Bedrock/Smithy HTTP/2 transport failures like http2 request did not get a response as transient errors, so the agent retries automatically instead of waiting for a manual nudge (#3594)
  • Fixed the CLI/SDK tool-selection split so --no-builtin-tools and createAgentSession({ noTools: "builtin" }) disable only built-in default tools while keeping extension/custom tools enabled, instead of falling through to the same "disable everything" path as --no-tools (#3592)
  • Fixed remaining hardcoded pi / .pi branding to route through APP_NAME and CONFIG_DIR_NAME extension points, so SDK rebrands get consistent naming in /quit description, process.title, and the project-local extensions directory (#3583 by @jlaneve)
  • Fixed pi-coding-agent shipping uuid@11, which triggered npm audit moderate vulnerability reports for downstream installs; the package now depends on uuid@14 (#3577)
  • Fixed openai-completions streamed tool-call assembly to coalesce deltas by stable tool index when OpenAI-compatible gateways mutate tool call IDs mid-stream, preventing malformed Kimi K2.6/OpenCode tool streams from splitting one call into multiple bogus tool calls (#3576)
  • Fixed ctx.ui.setWorkingMessage() to persist across loader recreation, matching the behavior of ctx.ui.setWorkingIndicator() (#3566)
  • Fixed coding-agent fs.watch error handling for theme and git-footer watchers to retry after transient watcher failures such as EMFILE, avoiding startup crashes in large repos (#3564)
  • Fixed built-in kimi-coding model generation to attach the expected User-Agent header so direct Kimi Coding requests use the provider's expected client identity (#3586)
  • Fixed extension shortcut conflict diagnostics to display at startup instead of only on reload, so extension authors discover reserved keybinding conflicts immediately rather than discovering them later through user feedback (#3617)
  • Fixed models.json Anthropic-compatible provider configuration to accept compat.supportsEagerToolInputStreaming, allowing proxies that reject per-tool eager_input_streaming to use the legacy fine-grained tool streaming beta header instead (#3575)
  • Fixed startup banner extension labels to strip trailing index.js/index.ts suffixes (#3596 by @aliou)
  • Fixed OSC 9;4 terminal progress updates to stay alive in terminals such as Ghostty during long-running agent work (#3610)
  • Fixed OpenAI-compatible completion usage parsing to avoid double-counting reasoning tokens already included in completion_tokens (#3581)
  • Fixed openai-responses compatibility for strict OpenAI-compatible proxies by allowing models.json to disable the underscore-containing session_id header with compat.sendSessionIdHeader: false (#3579)
  • Fixed GPT-5.5 Codex capability handling to clamp unsupported minimal reasoning to low and apply the model's 2.5x priority service-tier pricing multiplier (#3618 by @markusylisiurunen)
0.69.0 2026-04-22 新功能 4破坏性变更 2新增 3修复 5 TypeBox 1.x migration for extensions and SDK integrations, i

新功能4

  • TypeBox 1.x migration for extensions and SDK integrations, including TypeBox-native tool argument validation that now works in eval-restricted runtimes such as Cloudflare Workers. See docs/extensions.md and docs/sdk.md.
  • Stacked extension autocomplete providers via ctx.ui.addAutocompleteProvider(...), allowing extensions to layer custom completion logic on top of built-in slash and path completion. See docs/extensions.md#autocomplete-providers and examples/extensions/github-issue-autocomplete.ts.
  • Terminating tool results via terminate: true, allowing custom tools to end on a final tool call without paying for an automatic follow-up LLM turn. See docs/extensions.md and examples/extensions/structured-output.ts.
  • OSC 9;4 terminal progress indicators during agent streaming and compaction for supporting terminals.

破坏性变更2

  • Migrated first-party coding-agent code, SDK/examples/docs, and package metadata from @sinclair/typebox 0.34.x to typebox 1.x. New extensions, SDK integrations, and pi packages should depend on and import from typebox. Legacy extension loading still aliases the root @sinclair/typebox package, but @sinclair/typebox/compiler is no longer shimmed. This migration also picks up the new @mariozechner/pi-ai TypeBox-native validator path, so tool argument validation now works in eval-restricted runtimes such as Cloudflare Workers instead of being skipped (#3112)
  • Session-replacement commands now invalidate captured pre-replacement session-bound extension objects after ctx.newSession(), ctx.fork(), and ctx.switchSession(). Old pi and command ctx references now throw instead of silently targeting the replaced session. Migration: if code needs to keep working in the replacement session after one of those calls, pass withSession to that same method and do the post-switch work there. In practice, move post-switch pi.sendUserMessage(), pi.sendMessage(), and command-ctx/session-manager access into withSession, and use only the ReplacedSessionContext passed to that callback for session-bound operations. Footguns: withSession runs after the old extension instance has already received session_shutdown, old cleanup may already have invalidated captured state, captured old pi / old command ctx are stale, and previously extracted raw objects such as const sm = ctx.sessionManager remain the caller's responsibility and must not be reused after the switch.

新增3

  • Added support for terminating tool results via terminate: true, allowing custom tools to end the current tool batch without an automatic follow-up LLM call, plus a structured-output.ts extension example and extension docs showing the pattern (#3525)
  • Added OSC 9;4 terminal progress indicators during agent streaming and compaction, so terminals like iTerm2, WezTerm, Windows Terminal, and Kitty show activity in their tab bar
  • Added ctx.ui.addAutocompleteProvider(...) for stacking extension autocomplete providers on top of the built-in slash/path provider, plus a github-issue-autocomplete.ts example and extension docs (#2983)

修复5

  • Fixed exported session HTML to sanitize markdown link URLs before rendering them into anchor tags, blocking javascript:-style payloads while preserving safe links in shared/exported sessions (#3532)
  • Fixed ctx.getSystemPrompt() inside before_agent_start to reflect chained system-prompt changes made by earlier before_agent_start handlers, and clarified the extension docs around provider-payload rewrites and what ctx.getSystemPrompt() does and does not report (#3539)
  • Fixed built-in google-gemini-cli model lists and selector entries to include gemini-3.1-flash-lite-preview, so Cloud Code Assist users no longer need manual --model fallback selection to use it (#3545)
  • Fixed extension session-replacement flows so ctx.newSession(), ctx.fork(), ctx.switchSession(), and imported-session replacements fully rebind before post-switch work runs, added withSession replacement callbacks with fresh ReplacedSessionContext helpers, and make stale pre-replacement pi / ctx session-bound accesses throw instead of silently targeting the wrong session (#2860)
  • Fixed models.json built-in provider overrides to accept headers without requiring baseUrl, so request-header-only overrides now load and apply correctly (#3538)
0.68.1 2026-04-22 新功能 2新增 1修复 9 Fireworks provider support with built-in models and FIREWORK

新功能2

  • Fireworks provider support with built-in models and FIREWORKS_API_KEY auth. See README.md#providers--models and docs/providers.md.
  • Configurable inline tool image width via terminal.imageWidthCells in /settings. See docs/settings.md#terminal--images.

新增1

  • Added built-in Fireworks provider support, including FIREWORKS_API_KEY setup/docs and the default Fireworks model accounts/fireworks/models/kimi-k2p6 (#3519)

修复9

  • Fixed interactive inline tool images to honor configurable terminal.imageWidthCells via /settings, so tool-output images are no longer hard-capped to 60 terminal cells (#3508)
  • Fixed sessionDir in settings.json to expand ~, so portable session-directory settings no longer require a shell wrapper (#3514)
  • Fixed parallel tool-call rows to leave the pending state as soon as each tool is finalized, while still appending persisted tool results in assistant source order (#3503)
  • Fixed exported session markdown to render Markdown while showing HTML-like message content such as <file name="...">...</file> verbatim, so shared sessions match the TUI instead of letting the browser interpret message text (#3484)
  • Fixed exported session HTML to render grep and find output through their existing TUI renderers and ls output through a native template renderer, avoiding missing formatting and spacing artifacts in shared sessions (#3491 by @aliou)
  • Fixed @ autocomplete fuzzy search to follow symlinked directories and include symlinked paths in results (#3507)
  • Fixed proxied agent streams to preserve the proxy-safe serializable subset of stream options, including session, transport, retry-delay, metadata, header, cache-retention, and thinking-budget settings (#3512)
  • Hardened Anthropic streaming against malformed tool-call JSON by owning SSE parsing with defensive JSON repair, replacing the deprecated fine-grained-tool-streaming beta header with per-tool eager_input_streaming, and updating stale test model references (#3175)
  • Fixed Bedrock runtime endpoint resolution to stop pinning built-in regional endpoints over AWS_REGION / AWS_PROFILE, restoring us.* and eu.* inference profile support after v0.68.0 while preserving custom VPC/proxy endpoint overrides (#3481, #3485, #3486, #3487, #3488)
0.68.0 2026-04-20 新功能 4破坏性变更 3新增 7变更 2修复 24 Configurable streaming working indicator for extensions via

新功能4

  • Configurable streaming working indicator for extensions via ctx.ui.setWorkingIndicator(), including animated, static, and hidden indicators. See docs/tui.md#working-indicator, docs/extensions.md, and examples/extensions/working-indicator.ts.
  • before_agent_start now exposes systemPromptOptions (BuildSystemPromptOptions) so extensions can inspect the structured system-prompt inputs without re-discovering resources. See docs/extensions.md#before_agent_start and examples/extensions/prompt-customizer.ts.
  • Configurable keybindings for scoped model selector actions and session-tree filter actions. See docs/keybindings.md.
  • /clone duplicates the current active branch into a new session, while extensions can choose whether to fork before or at an entry via ctx.fork(..., { position }). See README.md, docs/extensions.md, and docs/session.md.

破坏性变更3

  • Changed SDK and CLI tool selection from cwd-bound built-in tool instances to tool-name allowlists. createAgentSession({ tools }) now expects string[] names such as "read" and "bash" instead of Tool[], --tools now allowlists built-in, extension, and custom tools by name, and --no-tools now disables all tools by default rather than only built-ins. Migrate SDK code from tools: [readTool, bashTool] to tools: ["read", "bash"] (#2835, #3452)
  • Removed prebuilt cwd-bound tool and tool-definition exports from @mariozechner/pi-coding-agent, including readTool, bashTool, editTool, writeTool, grepTool, findTool, lsTool, readOnlyTools, codingTools, and the corresponding *ToolDefinition values. Use the explicit factory exports instead, for example createReadTool(cwd), createBashTool(cwd), createCodingTools(cwd), and createReadToolDefinition(cwd) (#3452)
  • Removed ambient process.cwd() / default agent-dir fallback behavior from public resource helpers. DefaultResourceLoader, loadProjectContextFiles(), and loadSkills() now require explicit cwd/agent-dir style inputs, and exported system-prompt option types now require an explicit cwd. Pass the session or project cwd explicitly instead of relying on process-global defaults (#3452)

新增7

  • Added extension support for customizing the interactive streaming working indicator via ctx.ui.setWorkingIndicator(), including custom animated frames, static indicators, hidden indicators, a new working-indicator.ts example extension, and updated extension/TUI/RPC docs (#3413)
  • Added systemPromptOptions (BuildSystemPromptOptions) to before_agent_start extension events, so extensions can inspect the structured inputs used to build the current system prompt (#3473 by @dljsjr)
  • Added /clone to duplicate the current active branch into a new session, while keeping /fork focused on forking from a previous user message (#2962)
  • Added ctx.fork() support for position: "before" | "at" so extensions and integrations can branch before a user message or duplicate the current point in the conversation; the interactive clone/fork UX builds on that runtime support (#3431 by @mitsuhiko)
  • Added configurable keybinding ids for scoped model selector actions and tree filter actions, so those interactive shortcuts can be remapped in keybindings.json (#3343 by @mpazik)
  • Added PI_OAUTH_CALLBACK_HOST support for built-in OAuth login flows, allowing local callback servers used by pi auth to bind to a custom interface instead of hardcoded 127.0.0.1 (#3409 by @Michaelliv)
  • Added reason and targetSessionFile metadata to session_shutdown extension events, so extensions can distinguish quit, reload, new-session, resume, and fork teardown paths (#2863)

变更2

  • Changed pi update to batch npm package updates per scope and run git package updates with bounded parallelism, reducing multi-package update time while preserving skip behavior for pinned and already-current packages (#2980)
  • Changed Bedrock session requests to omit maxTokens when model token limits are unknown and to omit temperature when unset, letting Bedrock use provider defaults and avoid unnecessary TPM quota reservation (#3400 by @wirjo)

修复24

  • Fixed AgentSession system-prompt option initialization to avoid constructing an invalid empty BuildSystemPromptOptions, so npm run check passes after cwd became mandatory.
  • Fixed shell-path resolution to stop consulting ambient process.cwd() state during bash execution, so session/project-specific shellPath settings now follow the active coding-agent session cwd instead of the launcher cwd (#3452)
  • Fixed ctx.ui.setWorkingIndicator() custom frames to render verbatim instead of forcing the theme accent color, so extensions now own working-indicator coloring when they customize it (#3467)
  • Fixed pi update reinstalling npm packages that are already at the latest published version by checking the installed package version before running npm install <pkg>@latest (#3000)
  • Fixed @ autocomplete plain queries to stop matching against the full cwd/base path, so path fragments in worktree names no longer crowd out intended results such as @plan (#2778)
  • Fixed built-in tool wrapping to use the same extension-runner context path as extension tools, so built-in tools receive execution context and read can warn when the current model does not support images (#3429)
  • Fixed openai-completions assistant replay to preserve compat.requiresThinkingAsText text-part serialization, avoiding same-model follow-up crashes when previous assistant messages mix thinking and text (#3387)
  • Fixed direct OpenAI Chat Completions sessions to map sessionId and cacheRetention to prompt caching fields, sending prompt_cache_key when caching is enabled and prompt_cache_retention: "24h" for direct api.openai.com requests with long retention (#3426)
  • Fixed OpenAI-compatible Chat Completions sessions to optionally send aligned session_id, x-client-request-id, and x-session-affinity headers from sessionId via compat.sendSessionAffinityHeaders, improving cache-affinity routing for backends such as Fireworks (#3430)
  • Fixed threaded /resume session relationships and current-session detection to canonicalize symlinked session paths during selector comparisons, so shared session directories no longer break parent-child matching or active-session delete protection (#3364)
  • Fixed /session, Sessions docs, and CLI help to consistently document that session reuse supports both file paths and session IDs, and that /session shows the current session ID (#3390)
  • Fixed Windows pnpm global install detection to recognize \\.pnpm\\ store paths, so update notices now suggest pnpm install -g @mariozechner/pi-coding-agent instead of falling back to npm (#3378)
  • Fixed missing @sinclair/typebox runtime dependency in @mariozechner/pi-coding-agent, so strict pnpm installs no longer fail with ERR_MODULE_NOT_FOUND when starting pi (#3434)
  • Fixed xterm uppercase typing in the interactive editor by decoding printable modifyOtherKeys input and normalizing shifted letter matching, so Shift+letter no longer disappears in pi (#3436)
  • Fixed /compact to reuse the session thinking level for compaction summaries instead of forcing high, avoiding invalid reasoning-effort errors on github-copilot/claude-opus-4.7 sessions configured for medium thinking (#3438)
  • Fixed shared/exported plain-text tool output to preserve indentation instead of collapsing leading whitespace in the web share page (#3440)
  • Fixed exported share pages to use browser-safe T and O shortcuts with clickable header toggles for thinking and tool visibility instead of browser-reserved Ctrl+T / Ctrl+O bindings (#3374 by @vekexasia)
  • Fixed skill resolution to dedupe symlinked aliases by canonical path, so pi config no longer shows duplicate skill entries when ~/.pi/agent/skills points to ~/.agents/skills (#3417 by @rwachtler)
  • Fixed OpenRouter request attribution to include Pi app headers (HTTP-Referer: https://pi.dev, X-OpenRouter-Title: pi, X-OpenRouter-Categories: cli-agent) when sessions are created through the coding-agent SDK and install telemetry is enabled (#3414)
  • Fixed custom-model compat schema/docs to support cacheControlFormat: "anthropic" for OpenAI-compatible providers that expose Anthropic-style prompt caching via cache_control markers (#3392)
  • Fixed Cloud Code Assist tool schemas to strip JSON Schema meta-declaration keys before provider translation, avoiding validation failures for tool-enabled sessions that use $schema, $defs, and related metadata (#3412 by @vladlearns)
  • Fixed direct Bedrock sessions to honor model.baseUrl as the runtime client endpoint, restoring support for custom Bedrock VPC or proxy routes (#3402 by @wirjo)
  • Fixed the edit tool to coerce stringified edits JSON before validation, so models that send the array payload as a JSON string no longer fall back to ad-hoc shell edits (#3370 by @dannote)
  • Fixed package manifest positive glob entries to expand before loading packaged resources, restoring manifest patterns such as skills/**/*.md (#3350 by @neonspectra)
0.67.68 2026-04-17 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.67.67 2026-04-17 新功能 1新增 1修复 16 Bedrock sessions can now authenticate with AWS_BEARER_TOKEN_

新功能1

  • Bedrock sessions can now authenticate with AWS_BEARER_TOKEN_BEDROCK, enabling Converse API access without local SigV4 credentials. See docs/providers.md#amazon-bedrock.

新增1

  • Added Bedrock bearer-token authentication support via AWS_BEARER_TOKEN_BEDROCK, enabling coding-agent sessions to use Bedrock Converse without local SigV4 credentials (#3125 by @wirjo)

修复16

  • Fixed /scoped-models Alt+Up/Down to stay a no-op in the implicit all enabled state instead of materializing a full explicit enabled-model list and marking the selector dirty (#3331)
  • Fixed Mistral Small 4 default thinking requests to use the model's supported reasoning control, avoiding 400 errors when starting sessions on mistral-small-2603 and mistral-small-latest (#3338)
  • Fixed Qwen chat-template thinking replay to preserve prior thinking across turns, so affected OpenAI-compatible models keep multi-turn tool-call arguments instead of degrading to empty {} payloads (#3325)
  • Fixed exported HTML transcripts so text selection no longer triggers click-based expand/collapse toggles (#3332 by @xu0o0)
  • Fixed flaky git package update notifications by waiting for captured git command stdio to fully drain before comparing local and remote commit SHAs (#3027)
  • Fixed system prompt dates to use a stable YYYY-MM-DD format instead of locale-dependent output, keeping prompts deterministic across runtimes and locales (#2814)
  • Fixed auto-retry transient error detection to treat Network connection lost. as retryable, so dropped provider connections retry instead of terminating the agent (#3317)
  • Fixed compact interactive extension startup summaries to disambiguate package extensions and repeated local index.ts entries by using package-aware labels and the minimal parent path needed to make local entries unique (#3308)
  • Fixed git package dependency installation to use production installs (npm install --omit=dev) during both install and update flows, so extension runtime dependencies must come from dependencies and not devDependencies (#3009)
  • Fixed tool_result / afterToolCall extension handling for error results by forwarding details and isError overrides through AgentSession instead of dropping them when isError was already true (#3051)
  • Fixed missing root exports for RpcClient and RPC protocol types from @mariozechner/pi-coding-agent, so ESM consumers can import them from the main package entrypoint (#3275)
  • Fixed OpenAI Codex service-tier cost accounting to trust the explicitly requested tier when the API echoes the default tier in responses, keeping session cost displays aligned with the selected tier (#3307 by @markusylisiurunen)
  • Fixed parallel tool-call finalization to convert afterToolCall hook throws into error tool results instead of aborting the remaining tool batch (#3084)
  • Fixed Bun binary asset path resolution to honor PI_PACKAGE_DIR for built-in themes, HTML export templates, and interactive bundled assets (#3074)
  • Fixed user-message turn spacing in interactive mode by restoring an inter-message spacer before user turns (except the first user message), preventing assistant and user blocks from rendering flush together.
  • Fixed interactive /import handling to support quoted JSONL paths with spaces, route missing JSONL files through the non-fatal SessionImportFileNotFoundError path, and document the importFromJsonl() exceptions (SessionImportFileNotFoundError, MissingSessionCwdError).
0.67.6 2026-04-16 新功能 4新增 3变更 2修复 6 Prompt templates support an argument-hint frontmatter field

新功能4

  • Prompt templates support an argument-hint frontmatter field that renders before the description in the / autocomplete dropdown, using <angle> for required and [square] for optional arguments. See docs/prompt-templates.md#argument-hints.
  • New after_provider_response extension hook lets extensions inspect provider HTTP status codes and headers immediately after each response is received and before stream consumption begins. See docs/extensions.md.
  • Compact interactive startup header with a comma-separated view of loaded AGENTS.md files, prompt templates, skills, and extensions. Press Ctrl+O to toggle the expanded listing.
  • Markdown links in assistant output now render as OSC 8 hyperlinks on terminals that advertise support; unknown terminals and tmux/screen default to plain text so URLs are never silently dropped.

新增3

  • Added argument-hint frontmatter field for prompt templates, displayed before the description in the autocomplete dropdown (#2780 by @andresvi94)
  • Added after_provider_response extension hook so extensions can inspect provider HTTP status codes and headers after each provider response is received and before stream consumption begins (#3128)
  • Added OSC 8 hyperlink rendering for markdown links when the terminal advertises support (#3248 by @ofa1)

变更2

  • Changed interactive startup header to a compact, comma-separated view of loaded AGENTS.md files, prompt templates, skills, and extensions, with Ctrl+O to toggle the expanded listing (#3267)
  • Tightened hyperlink capability detection to default hyperlinks: false for unknown terminals and force it off under tmux/screen (including nested sessions), preventing markdown link URLs from disappearing on terminals that silently swallow OSC 8 sequences (#3248)

修复6

  • Fixed interactive user message rendering to keep bottom padding visible in terminals affected by OSC 133 prompt markers without adding an extra blank line before the following assistant message (#3090)
  • Fixed --verbose startup output to begin with expanded startup help and loaded resource listings after the compact startup header change (#3147)
  • Fixed find tool returning no results for path-based glob patterns such as src/**/*.spec.ts or some/parent/child/** by switching fd into full-path mode and normalizing the pattern when it contains a / (#3302)
  • Fixed find tool applying nested .gitignore rules across sibling directories (e.g. rules from a/.gitignore hiding matching files under b/) by dropping the manual --ignore-file collection and delegating to fd's hierarchical .gitignore handling via --no-require-git (#3303)
  • Fixed OpenAI Responses prompt caching for non-api.openai.com base URLs (OpenAI-compatible proxies such as litellm, theclawbay) by sending the session_id and x-client-request-id cache-affinity headers unconditionally when a sessionId is provided, matching the official Codex CLI behavior (#3264 by @vegarsti)
  • Fixed the preset example extension to snapshot the active model, thinking level, and tool set on the first preset application and restore that state when cycling back to (none), instead of falling back to a hardcoded default tool list (#3272 by @stembi)
0.67.5 2026-04-16 修复 2 Fixed Opus 4.7 adaptive thinking configuration across Anthro

修复2

  • Fixed Opus 4.7 adaptive thinking configuration across Anthropic and Bedrock providers by recognizing Opus 4.7 adaptive-thinking support and mapping xhigh reasoning to provider-supported effort values (#3286 by @markusylisiurunen)
  • Fixed Zellij Shift+Enter regressions by reverting the Zellij-specific Kitty keyboard query bypass and restoring the previous keyboard negotiation behavior (#3259)
0.67.4 2026-04-16 新功能 3新增 3变更 2修复 7 --no-context-files (-nc) disables automatic AGENTS.md / CLAU

新功能3

  • --no-context-files (-nc) disables automatic AGENTS.md / CLAUDE.md discovery when you need a clean run without project context injection. See README.md#context-files.
  • loadProjectContextFiles() is now exported as a standalone utility for extensions and SDK-style integrations that need to inspect the same context-file resolution order used by the CLI. See README.md#context-files.
  • New after_provider_response extension hook lets extensions inspect provider HTTP status codes and headers immediately after response creation and before stream consumption. See docs/extensions.md.

新增3

  • Added --no-context-files (-nc) to disable AGENTS.md and CLAUDE.md context file discovery and loading (#3253)
  • Exported loadProjectContextFiles() as a standalone utility so extensions can discover project context files without instantiating a full DefaultResourceLoader (#3142)
  • Added after_provider_response extension hook so extensions can inspect provider HTTP status codes and headers after each provider response is received and before stream consumption begins (#3128)

变更2

  • Added claude-opus-4-7 model for Anthropic.
  • Changed Anthropic prompt caching to add a cache_control breakpoint on the last tool definition, so tool schemas can be cached independently from transcript updates while preserving existing cache retention behavior (#3260)

修复7

  • Fixed markdown strikethrough parsing in interactive rendering and HTML export to require strict double-tilde delimiters (~~text~~) with non-whitespace boundaries.
  • Fixed shutdown handling to kill tracked detached bash tool child processes on exit signals, preventing orphaned background processes.
  • Fixed flaky edit-tool-no-full-redraw TUI tests by waiting for asynchronous preview and preflight error rendering instead of relying on fixed render ticks.
  • Fixed kimi-coding default model selection to use kimi-for-coding instead of kimi-k2-thinking (#3242)
  • Fixed ctrl+z on native Windows to avoid crashing interactive mode, disable the default suspend binding there, and show a status message when suspend is invoked manually (#3191)
  • Fixed find tool cancellation and responsiveness on broad searches by making .gitignore discovery and fd execution fully abort-aware and non-blocking (#3148)
  • Fixed grep broad-search stalls when context=0 by formatting match lines from ripgrep JSON output instead of doing synchronous per-match file reads (#3205)
0.67.3 2026-04-15 新功能 2新增 1修复 9 renderShell: "self" for custom and built-in tool renderers s

新功能2

  • renderShell: "self" for custom and built-in tool renderers so tools can own their outer shell instead of the default boxed shell. Useful for stable large previews such as edit diffs. See docs/extensions.md#custom-rendering.
  • Interactive auto-retry status now shows a live countdown during backoff instead of a static retry delay message.

新增1

  • Added renderShell: "self" for custom and built-in tool renderers so tools can own their outer shell instead of using the default boxed shell. This is useful for stable large previews such as edit diffs (#3134)

修复9

  • Fixed edit diff previews to stay visible during edit permission dialogs and session replay without reintroducing large-result redraw flicker (#3134)
  • Fixed /reload to render a static reload status box instead of an animated spinner, avoiding redraw instability during interactive reloads.
  • Fixed the plan-mode example extension to allow eza in the read-only bash allowlist instead of the deprecated exa command (#3240 by @rwachtler)
  • Fixed google-vertex API key resolution to treat gcp-vertex-credentials as an Application Default Credentials marker instead of a literal API key, so marker-based setups correctly fall back to ADC (#3221 by @deepkilo)
  • Fixed RPC prompt to wait for prompt preflight success before emitting its single authoritative response, while still treating handled and queued prompts as success (#3049)
  • Fixed /scoped-models reordering to propagate into the /model scoped tab, preserving the user-defined scoped model order instead of re-sorting it (#3217)
  • Fixed session_shutdown to fire on SIGHUP and SIGTERM in interactive, print, and RPC modes so extensions can run shutdown cleanup on those signal-driven exits (#3212)
  • Fixed screenshot path parsing to handle lower case am/pm in macOS screenshot filenames (#3194 by @jay-aye-see-kay)
  • Fixed interactive auto-retry status updates to show a live countdown during backoff instead of a static retry delay message (#3187)
0.67.2 2026-04-14 新功能 3新增 3修复 5 Support for multiple --append-system-prompt flags, each valu

新功能3

  • Support for multiple --append-system-prompt flags, each value is appended to the system prompt separated by double newlines. See README.md#other-options.
  • Support for passing inline extension factories to main() for embedded integrations and custom entrypoints.
  • Interactive keybinding support for Kitty super-modified shortcuts such as super+k, super+enter, and ctrl+super+k. See docs/keybindings.md.

新增3

  • Added support for multiple --append-system-prompt flags, each value is appended to the system prompt separated by double newlines (#3171 by @aliou)
  • Added interactive keybinding support for Kitty super-modified shortcuts such as super+k, super+enter, and ctrl+super+k (#3111 by @sudosubin)
  • Added support for passing inline extension factories to main() for embedded integrations and custom entrypoints (#3099 by @pmateusz)

修复5

  • Fixed direct OpenAI Responses and Codex SSE requests to align prompt_cache_key, session_id, and x-client-request-id values with the same session-derived identifier, improving prompt cache affinity for append-only sessions (#3018 by @steipete)
  • Fixed streaming-only partialJson scratch buffers leaking into persisted OpenAI Responses tool calls, which could corrupt follow-up payloads on resumed conversations.
  • Fixed Ctrl+Alt letter key matching in tmux by falling through from legacy ESC-prefixed handling to CSI-u and xterm modifyOtherKeys parsing when the legacy form does not match (#2989 by @kaofelix)
  • Fixed the shipped subagent example to avoid leaking Bun virtual filesystem script paths into subagent prompts (#3002 by @nathyong)
  • Fixed bordered loaders to stop their animation timer when disposed, preventing stale loader updates after teardown.
0.67.1 2026-04-13 其他 14新功能 5新增 2修复 11 Pi needs a reliable per-version usage signal to understand w

Telemetry14

  • Pi needs a reliable per-version usage signal to understand whether releases are being adopted and to help justify funding continued development.
  • npm download counts are not a reliable proxy for actual Pi usage.
  • It only runs in interactive mode.
  • It does not run in RPC mode, print mode, JSON mode, or SDK mode.
  • On a fresh interactive install, Pi writes lastChangelogVersion, then sends the ping.
  • On later interactive startups, if the local changelog contains entries newer than the previously stored lastChangelogVersion, Pi writes the new lastChangelogVersion, then sends the ping.
  • The request is fire-and-forget. Startup does not wait for it, and any errors are ignored.
  • Only the Pi version in the request path, for example https://pi.dev/install?version=0.67.1.
  • The server stores only aggregate per-version counters such as { "0.67.1": 3 }.
  • It does not store IP addresses, client identifiers, prompts, paths, models, auth state, or any other per-user data. It literally only increments a counter for that version.
  • /settings → disable Install telemetry
  • settings.json → set enableInstallTelemetry to false
  • PI_OFFLINE=1
  • PI_TELEMETRY=0

新功能5

  • Full openRouterRouting support in models.json, including fallbacks, parameter requirements, data collection, ZDR, ignore lists, quantizations, provider sorting, max price, and preferred throughput and latency constraints. See docs/models.md.
  • PI_CODING_AGENT=true environment variable set at startup so subprocesses can detect they are running inside the coding agent.
  • Updated antigravity-image-gen.ts example extension to use User-Agent version 1.21.9 (#2901 by @aadishv)
  • Fixed --list-models silently swallowing models.json load errors; errors are now printed to stderr (#3072)
  • Fixed custom models for built-in providers (e.g. openrouter) being silently dropped from --list-models by inheriting api/baseUrl from built-in model definitions and no longer requiring apiKey for providers with existing auth (#2921 and #3072)

新增2

  • Added full openRouterRouting field support in models.json, including fallbacks, parameter requirements, data collection, ZDR, ignore lists, quantizations, provider sorting, max price, and preferred throughput and latency constraints (#2904 by @zmberber)
  • Set PI_CODING_AGENT=true environment variable at startup so sub-processes can detect they are running inside the coding agent (#2868)

修复11

  • Fixed interactive changelog rendering for the telemetry notes by moving the section under a ### Telemetry heading, so startup shows the full release notes instead of only the version header.
  • Updated antigravity-image-gen.ts example extension to use User-Agent version 1.21.9 (#2901 by @aadishv)
  • Bumped default Antigravity User-Agent version to 1.21.9 (#2901 by @aadishv)
  • Fixed Gemma 4 thinking level mapping to route between MINIMAL and HIGH, and map Pi reasoning levels to the model's supported thinking levels (#2903 by @aadishv)
  • Fixed Gemini 2.5 Flash Lite minimal thinking budget to use the model's supported 512-token minimum instead of the regular Flash 128-token minimum, avoiding invalid thinking budget errors (#2861 by @JasonOA888)
  • Fixed OpenAI Codex Responses requests to forward configured serviceTier values, restoring service-tier selection for Codex sessions (#2996 by @markusylisiurunen)
  • Fixed newly generated session IDs to use UUIDv7, improving time locality for session-based request routing (#3018 by @steipete)
  • Fixed Container.render() stack overflow on long sessions by replacing Array.push(...spread) with a loop-based push, preventing RangeError: Maximum call stack size exceeded when child output exceeds the V8 call stack argument limit (#2651)
  • Fixed editor sticky-column tracking around paste markers so vertical cursor navigation restores the column from before the cursor entered a paste marker instead of jumping inside or past pasted content (#3092 by @Perlence)
  • Fixed queued messages typed during /tree branch summarization to flush automatically after navigation completes, so they no longer remain stuck in the steering queue (#3091 by @Perlence)
  • Fixed npm package update check to work with packages on non-default registries by using npm view instead of hardcoded registry.npmjs.org fetch (#3164 by @aliou)
0.67.0 2026-04-13 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.66.1 2026-04-08 变更 1 Changed the Earendil announcement from an automatic startup

变更1

  • Changed the Earendil announcement from an automatic startup notice to the hidden /dementedelves slash command.
0.66.0 2026-04-08 新功能 2修复 4 Earendil startup announcement with bundled inline image rend

新功能2

  • Earendil startup announcement with bundled inline image rendering and a linked blog post for April 8 and 9, 2026.
  • Interactive Anthropic subscription auth warning when Anthropic subscription auth is active, clarifying that Anthropic third-party usage draws from extra usage and is billed per token.

修复4

  • Fixed bare readline import to use node:readline prefix for Deno compatibility (#2885 by @milosv-vtool)
  • Fixed auto-retry to treat stream failures like request ended without sending any chunks as transient errors (#2892)
  • Fixed interactive startup notices to render after the initial resource listing, and added a bundled Earendil startup announcement with inline image rendering for April 8 and 9, 2026. Moved the blog link above the image to avoid overlap with terminal image rendering.
  • Fixed interactive mode to warn when Anthropic subscription auth is active, so users know Anthropic third-party usage draws from extra usage and is billed per token.
0.65.2 2026-04-06 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.65.1 2026-04-05 修复 9 Fixed bash output truncation by line count to always persist

修复9

  • Fixed bash output truncation by line count to always persist full output to a temp file, preventing data loss when output exceeds 2000 lines but stays under the byte threshold (#2852)
  • RpcClient now forwards subprocess stderr to parent process in real-time (#2805)
  • Theme file watcher now handles async fs.watch error events instead of crashing the process (#2791)
  • Fixed stored session cwd handling so resuming or importing a session whose original working directory no longer exists now prompts interactive users to continue in the current cwd, while non-interactive modes fail with a clear error.
  • Fixed resource collision precedence so project and user skills, prompt templates, and themes override package resources consistently, and CLI-provided paths take precedence over discovered resources (#2781)
  • Fixed OpenAI-compatible completions streaming usage accounting to preserve prompt_tokens_details.cache_write_tokens and normalize OpenRouter cached_tokens, preventing incorrect cache read/write token and cost reporting in pi (#2802)
  • Fixed CLI extension paths like git:gist.github.com/... being incorrectly resolved against cwd instead of being passed through to the package manager (#2845 by @aliou)
  • Fixed piped stdin runs with --mode json to preserve JSONL output instead of falling back to plain text (#2848 by @aliou)
  • Fixed interactive command docs to stop listing removed /exit as a supported quit command (#2850)
0.65.0 2026-04-03 新功能 4破坏性变更 4新增 5修复 5 Session runtime API

新功能4

  • Session runtime API: createAgentSessionRuntime() and AgentSessionRuntime provide a closure-based runtime that recreates cwd-bound services and session config on every session switch. Startup, /new, /resume, /fork, and import all use the same creation path. See docs/sdk.md and examples/sdk/13-session-runtime.ts.
  • Label timestamps in /tree: Toggle timestamps on tree entries with Shift+T, with smart date formatting and timestamp preservation through branching (#2691 by @w-winter)
  • defineTool() helper: Create standalone custom tool definitions with full TypeScript parameter type inference, no manual casts needed (#2746). See docs/extensions.md.
  • Unified diagnostics: Arg parsing, service creation, session option resolution, and resource loading all return structured diagnostics (info/warning/error) instead of logging or exiting. The app layer decides presentation and exit behavior.

破坏性变更4

  • Removed extension post-transition events session_switch and session_fork. Use session_start with event.reason ("startup" | "reload" | "new" | "resume" | "fork"). For "new", "resume", and "fork", session_start includes previousSessionFile.
  • Removed session-replacement methods from AgentSession. Use AgentSessionRuntime for newSession(), switchSession(), fork(), and importFromJsonl(). Cross-cwd session replacement rebuilds all cwd-bound runtime state and replaces the live AgentSession instance.
  • Removed session_directory from extension and settings APIs.
  • Unknown single-dash CLI flags (e.g. -s) now produce an error instead of being silently ignored.

    // event.reason: "startup" | "reload" | "new" | "resume" | "fork"

    // event.previousSessionFile: set for "new", "resume", "fork"

    });

    #### Migration: SDK session replacement
Before:

    await session.newSession();

    await session.switchSession("/path/to/session.jsonl");

    After:

    import {

    type CreateAgentSessionRuntimeFactory,

    createAgentSessionFromServices,

    createAgentSessionRuntime,

    createAgentSessionServices,

    getAgentDir,

    SessionManager,

    } from "@mariozechner/pi-coding-agent";

    const createRuntime: CreateAgentSessionRuntimeFactory = async ({ cwd, sessionManager, sessionStartEvent }) => {

    const services = await createAgentSessionServices({ cwd });

    return {

    ...(await createAgentSessionFromServices({ services, sessionManager, sessionStartEvent })),

    services,

    diagnostics: services.diagnostics,

    };

    };

    const runtime = await createAgentSessionRuntime(createRuntime, {

    cwd: process.cwd(),

    agentDir: getAgentDir(),

    sessionManager: SessionManager.create(process.cwd()),

    });

    await runtime.newSession();

    await runtime.switchSession("/path/to/session.jsonl");

    await runtime.fork("entry-id");

    // After replacement, runtime.session is the new live session.

    // Rebind any session-local subscriptions or extension bindings.

新增5

  • Added createAgentSessionRuntime() and AgentSessionRuntime for runtime-backed session replacement. The runtime takes a CreateAgentSessionRuntimeFactory closure that closes over process-global fixed inputs and recreates cwd-bound services and session config for each effective cwd. Startup and later /new, /resume, /fork, import all use the same factory.
  • Added unified diagnostics model (info/warning/error) for arg parsing, service creation, session option resolution, and resource loading. Creation logic no longer logs or exits. The app layer decides presentation and exit behavior.
  • Added error diagnostics for missing explicit CLI resource paths (-e, --skill, --prompt-template, --theme)
  • Added defineTool() so standalone and array-based custom tool definitions keep inferred parameter types without manual casts (#2746)
  • Added label timestamps to the session tree with a Shift+T toggle in /tree, smart date formatting, and timestamp preservation through branching (#2691 by @w-winter)

修复5

  • Fixed startup resource loading to reuse the initial ResourceLoader for the first runtime, so extensions are not loaded twice before session startup and session_start handlers still fire for singleton-style extensions (#2766)
  • Fixed retry settlement so retried agent runs wait for the full retry cycle to complete before declaring idle, preventing stale state after transient errors
  • Fixed theme export colors to resolve theme variables the same way as colors, so /export HTML backgrounds now honor entries like pageBg: "base" instead of requiring inline hex values (#2707)
  • Fixed Bedrock throttling errors being misidentified as context overflow, causing unnecessary compaction instead of retry (#2699 by @xu0o0)
  • Added tool streaming support for newer Z.ai models (#2732 by @kaofelix)
2026-0321 个版本
0.64.0 2026-03-29 新功能 2破坏性变更 1新增 3修复 4 Extensions and SDK callers can attach a prepareArguments hoo

新功能2

  • Extensions and SDK callers can attach a prepareArguments hook to any tool definition, letting them normalize or migrate raw model arguments before schema validation. The built-in edit tool uses this to transparently support sessions created with the old single-edit schema. See docs/extensions.md
  • Extensions can customize the collapsed thinking block label via ctx.ui.setHiddenThinkingLabel(). See examples/extensions/hidden-thinking-label.ts (#2673)

破坏性变更1

  • ModelRegistry no longer has a public constructor. SDK callers and tests must use ModelRegistry.create(authStorage, modelsJsonPath?) for file-backed registries or ModelRegistry.inMemory(authStorage) for built-in-only registries. Direct new ModelRegistry(...) calls no longer compile.

新增3

  • Added ToolDefinition.prepareArguments hook to prepare raw tool call arguments before schema validation, enabling compatibility shims for resumed sessions with outdated tool schemas
  • Built-in edit tool now uses prepareArguments to silently fold legacy top-level oldText/newText into edits[] when resuming old sessions
  • Added ctx.ui.setHiddenThinkingLabel() so extensions can customize the collapsed thinking label in interactive mode, with a no-op in RPC mode and a runnable example extension in examples/extensions/hidden-thinking-label.ts (#2673)

修复4

  • Fixed extension-queued user messages to refresh the interactive pending-message list so messages submitted while a turn is active are no longer silently dropped (#2674 by @mrexodia)
  • Fixed monorepo tsconfig.json path mappings to resolve @mariozechner/pi-ai subpath exports to source files in development checkouts (#2625 by @ferologics)
  • Fixed TUI cell size response handling to consume only exact CSI 6 ; height ; width t replies, so bare Escape is no longer swallowed while waiting for terminal image metadata (#2661)
  • Fixed Kitty keyboard protocol keypad functional keys to normalize to logical digits, symbols, and navigation keys, so numpad input in terminals such as iTerm2 no longer inserts Private Use Area gibberish or gets ignored (#2650)
0.63.2 2026-03-29 新功能 3新增 1修复 2 Extension handlers can now use ctx.signal to forward cancell

新功能3

  • Extension handlers can now use ctx.signal to forward cancellation into nested model calls, fetch(), and other abort-aware work. See docs/extensions.md#ctxsignal (#2660)
  • Built-in edit tool input now uses edits[] as the only replacement shape, reducing invalid tool calls caused by mixed single-edit and multi-edit schemas (#2639)
  • Large multi-edit results no longer trigger full-screen redraws in the interactive TUI when the final diff is rendered (#2664)

新增1

  • Added ctx.signal to ExtensionContext and wired it to the active agent turn so extension handlers can forward cancellation into nested model calls, fetch(), and other abort-aware work (#2660)

修复2

  • Fixed built-in edit tool input to use edits[] as the only replacement shape, eliminating the mixed single-edit and multi-edit modes that caused repeated invalid tool calls and retries (#2639)
  • Fixed edit tool TUI rendering to defer large multi-edit diffs to the settled result, avoiding full-screen redraws when the tool completes (#2664)
0.63.1 2026-03-27 新增 1修复 9 Added gemini-3.1-pro-preview-customtools model availability

新增1

  • Added gemini-3.1-pro-preview-customtools model availability for the google-vertex provider (#2610 by @gordonhwc)

修复9

  • Documented tool_call input mutation as supported extension API behavior, clarified that post-mutation inputs are not re-validated, and added regression coverage for executing mutated tool arguments (#2611)
  • Fixed repeated compactions dropping messages that were kept by an earlier compaction by re-summarizing from the previous kept boundary and recalculating tokensBefore from the rebuilt session context (#2608)
  • Fixed interactive compaction UI updates so ctx.compact() rebuilds the chat through unified compaction events, manual compaction no longer duplicates the summary block, and the trigger-compact example only fires when context usage crosses its threshold (#2617)
  • Fixed interactive compaction completion to append a synthetic compaction summary after rebuilding the chat so the latest compaction remains visible at the bottom
  • Fixed skill discovery to stop recursing once a directory contains SKILL.md, and to ignore root *.md files in .agents/skills while keeping root markdown skill files supported in ~/.pi/agent/skills, .pi/skills, and package skills/ directories (#2603)
  • Fixed edit tool diff rendering for multi-edit operations with large unchanged gaps so distant edits collapse intermediate context instead of dumping the full unchanged middle block
  • Fixed edit tool error rendering to avoid repeating the same exact-match failure in both the preview and result blocks
  • Fixed auto-compaction overflow recovery for Ollama models when the backend returns explicit prompt too long; exceeded max context length ... errors instead of silently truncating input (#2626)
  • Fixed built-in tool overrides that reuse built-in parameter schemas to still honor custom renderCall and renderResult renderers in the interactive TUI, restoring the minimal-mode example (#2595)
0.63.0 2026-03-27 破坏性变更 2新增 4修复 14 ModelRegistry.getApiKey(model) has been replaced by getApiKe

破坏性变更2

  • ModelRegistry.getApiKey(model) has been replaced by getApiKeyAndHeaders(model) because models.json auth and header values can now resolve dynamically on every request. Extensions and SDK integrations that previously fetched only an API key must now fetch request auth per call and forward both apiKey and headers. Use getApiKeyForProvider(provider) only when you explicitly want provider-level API key lookup without model headers or authHeader handling (#1835)
  • Removed deprecated direct minimax and minimax-cn model IDs, keeping only MiniMax-M2.7 and MiniMax-M2.7-highspeed. Update pinned model IDs to one of those supported direct MiniMax models, or use another provider route that still exposes the older IDs (#2596 by @liyuan97)

    apiKey: auth.apiKey,

    headers: auth.headers,

    });

新增4

  • Added sessionDir setting support in global and project settings.json so session storage can be configured without passing --session-dir on every invocation (#2598 by @smcllns)
  • Added a startup onboarding hint in the interactive header telling users pi can explain its own features and documentation (#2620 by @ferologics)
  • Added edit tool multi-edit support so one call can update multiple separate, disjoint regions in the same file while matching all replacements against the original file content
  • Added support for PI_TUI_WRITE_LOG directory paths, creating a unique log file (tui-<timestamp>-<pid>.log) per instance for easier debugging of multiple pi sessions (#2508 by @mrexodia)

修复14

  • Fixed file mutation queue ordering so concurrent edit and write operations targeting the same file stay serialized in request order instead of being reordered during queue-key resolution
  • Fixed models.json shell-command auth and headers to resolve at request time instead of being cached into long-lived model state. pi now leaves TTL, caching, and recovery policy to user-provided wrapper commands because arbitrary shell commands need provider-specific strategies (#1835)
  • Fixed Google and Vertex cost calculation to subtract cached prompt tokens from billable input tokens instead of double-counting them when providers report cachedContentTokenCount (#2588 by @sparkleMing)
  • Added missing ajv direct dependency; previously relied on transitive install via @mariozechner/pi-ai which broke standalone installs (#2252)
  • Fixed /export HTML backgrounds to honor theme.export.pageBg, cardBg, and infoBg instead of always deriving them from userMessageBg (#2565)
  • Fixed interactive bash execution collapsed previews to recompute visual line wrapping at render time, so previews respect the current terminal width after resizes and split-pane width changes (#2569)
  • Fixed RPC get_session_stats to expose contextUsage, so headless clients can read actual current context-window usage instead of deriving it from token totals (#2550)
  • Fixed pi update for git packages to fetch only the tracked target branch with --no-tags, reducing unrelated branch and tag noise while preserving force-push-safe updates (#2548)
  • Fixed print and JSON modes to emit session_shutdown before exit, so extensions can release long-lived resources and non-interactive runs terminate cleanly (#2576)
  • Fixed GitHub Copilot OpenAI Responses requests to omit the reasoning field entirely when no reasoning effort is requested, avoiding 400 errors from Copilot gpt-5-mini rejecting reasoning: { effort: "none" } during internal summary calls (#2567)
  • Fixed blockquote text color breaking after inline links (and other inline elements) due to missing style restoration prefix
  • Fixed slash-command Tab completion from immediately chaining into argument autocomplete after completing the command name, restoring flows like /model that submit into a selector dialog (#2577)
  • Fixed stale content and incorrect viewport tracking after TUI content shrinks or transient components inflate the working area (#2126 by @Perlence)
  • Fixed @ autocomplete to debounce editor-triggered searches, cancel in-flight fd lookups cleanly, and keep suggestions visible while results refresh (#1278)
0.62.0 2026-03-23 新功能 3破坏性变更 14变更 3修复 12 Built-in tools as extensible ToolDefinitions. Extension auth

新功能3

  • Built-in tools as extensible ToolDefinitions. Extension authors can now override rendering of built-in read/write/edit/bash/grep/find/ls tools with custom renderCall/renderResult components. See docs/extensions.md.
  • Unified source provenance via sourceInfo. All resources, commands, tools, skills, and prompt templates now carry structured sourceInfo with path, scope, and source metadata. Visible in autocomplete, RPC discovery, and SDK introspection. See docs/extensions.md.
  • AWS Bedrock cost allocation tagging. New requestMetadata option on BedrockOptions forwards key-value pairs to the Bedrock Converse API for AWS Cost Explorer split cost allocation.

破坏性变更14

  • Changed ToolDefinition.renderCall and renderResult semantics. Fallback rendering now happens only when a renderer is not defined for that slot. If renderCall or renderResult is defined, it must return a Component.
  • Changed slash command provenance to use sourceInfo consistently. RPC get_commands, RpcSlashCommand, and SDK SlashCommandInfo no longer expose location or path. Use sourceInfo instead (#1734)
  • Removed legacy source fields from Skill and PromptTemplate. Use sourceInfo.source for provenance instead (#1734)
  • Removed ResourceLoader.getPathMetadata(). Resource provenance is now attached directly to loaded resources via sourceInfo (#1734)
  • Removed extensionPath from RegisteredCommand and RegisteredTool. Use sourceInfo.path for provenance instead (#1734)
  • RPC get_commands: replace path and location with sourceInfo.path, sourceInfo.scope, and sourceInfo.source
  • SlashCommandInfo: replace command.path and command.location with command.sourceInfo
  • Skill and PromptTemplate: replace .source with .sourceInfo.source
  • RegisteredCommand and RegisteredTool: replace .extensionPath with .sourceInfo.path
  • Custom ResourceLoader implementations: remove getPathMetadata() and read provenance from loaded resources directly
  • command.path -> command.sourceInfo.path
  • command.location === "user" -> command.sourceInfo.scope === "user"
  • skill.source -> skill.sourceInfo.source
  • tool.extensionPath -> tool.sourceInfo.path

变更3

  • Built-in tools now work like custom tools in extensions. To get built-in tool definitions, import readToolDefinition / createReadToolDefinition() and the equivalent bash, edit, write, grep, find, and ls exports from @mariozechner/pi-coding-agent.
  • Cleaned up buildSystemPrompt() so built-in tool snippets and tool-local guidelines come from built-in ToolDefinition metadata, while cross-tool and global prompt rules stay in system prompt construction.
  • Added structured sourceInfo to pi.getAllTools() results for built-in, SDK, and extension tools (#1734)

修复12

  • Fixed extension command name conflicts so extensions with duplicate command names can load together. Conflicting extension commands now get numeric invocation suffixes in load order, for example /review:1 and /review:2 (#1061)
  • Fixed slash command source attribution for extension commands, prompt templates, and skills in autocomplete and command discovery (#1734)
  • Fixed auto-resized image handling to enforce the inline image size limit on the final base64 payload, return text-only fallbacks when resizing cannot produce a safe image, and avoid falling back to the original image in read and @file auto-resize paths (#2055)
  • Fixed pi update for git packages to skip destructive reset, clean, and reinstall steps when the fetched target already matches the local checkout (#2503)
  • Fixed print and JSON mode to take over stdout during non-interactive startup, keeping package-manager and other incidental chatter off protocol/output stdout (#2482)
  • Fixed cli-highlight auto-detection for languageless code blocks that misidentified prose as programming languages and colored random English words as keywords
  • Fixed Anthropic thinking disable handling to send thinking: { type: "disabled" } for reasoning-capable models when thinking is explicitly off (#2022)
  • Fixed explicit thinking disable handling across Google, Google Vertex, Gemini CLI, OpenAI Responses, Azure OpenAI Responses, and OpenRouter-backed OpenAI-compatible completions (#2490)
  • Fixed OpenAI Responses replay for foreign tool-call item IDs by hashing foreign IDs into bounded fc_<hash> IDs
  • Fixed OpenAI-compatible completions streams to ignore null chunks instead of crashing (#2466 by @Cheng-Zi-Qing)
  • Fixed truncateToWidth() performance for very large strings by streaming truncation (#2447)
  • Fixed markdown heading styling being lost after inline code spans within headings
0.61.1 2026-03-20 新功能 2新增 1变更 1修复 5 Typed tool_call handler return values via ToolCallEventResul

新功能2

  • Typed tool_call handler return values via ToolCallEventResult exports from the top-level package and core extension entry. See docs/extensions.md.
  • Updated default models for zai, cerebras, minimax, and minimax-cn, and aligned MiniMax catalog coverage and limits with the current provider lineup. See docs/models.md and docs/providers.md.

新增1

  • Added ToolCallEventResult to the @mariozechner/pi-coding-agent top-level and core extension exports so extension authors can type explicit tool_call handler return values (#2458)

变更1

  • Changed the default models for zai, cerebras, minimax, and minimax-cn to match the current provider lineup, and added missing MiniMax-M2.1-highspeed model entries with normalized MiniMax context limits (#2445 by @1500256797)

修复5

  • Fixed ctrl+z suspend and fg resume reliability by keeping the process alive until the SIGCONT handler restores the TUI, avoiding immediate process exit in environments with no other live event-loop handles (#2454)
  • Fixed createAgentSession({ agentDir }) to derive the default persisted session path from the provided agentDir, keeping session storage aligned with settings, auth, models, and resource loading (#2457)
  • Fixed shared keybinding resolution to stop user overrides from evicting unrelated default shortcuts such as selector confirm and editor cursor keys (#2455)
  • Fixed Termux software keyboard height changes from forcing full-screen redraws and replaying TUI history on every toggle (#2467)
  • Fixed project-local npm package updates to install npm latest instead of reusing stale saved dependency ranges, and added Did you mean ...? suggestions when pi update <source> omits the configured npm or git source prefix (#2459)
0.61.0 2026-03-20 新功能 3破坏性变更 2新增 3修复 18 Namespaced keybinding ids and a unified keybinding manager a

新功能3

  • Namespaced keybinding ids and a unified keybinding manager across the app and TUI. See docs/keybindings.md and docs/extensions.md.
  • JSONL session export and import via /export <path.jsonl> and /import <path.jsonl>. See README.md and docs/session.md.
  • Resizable sidebar in HTML share and export views. See README.md.

破坏性变更2

  • Interactive keybinding ids are now namespaced, and keybindings.json now uses those same canonical namespaced ids. Older config files are migrated automatically on startup. Custom editors and extension UI components still receive an injected keybindings: KeybindingsManager. They do not call getKeybindings() or setKeybindings() themselves. Declaration merging applies to that injected type (#2391)
  • Extension author migration: update keyHint(), keyText(), and injected keybindings.matches(...) calls from old built-in names like "expandTools", "selectConfirm", and "interrupt" to namespaced ids like "app.tools.expand", "tui.select.confirm", and "app.interrupt". See docs/keybindings.md for the full list. pi.registerShortcut("ctrl+shift+p", ...) is unchanged because extension shortcuts still use raw key combos, not keybinding ids.

新增3

  • Added gpt-5.4-mini to the openai-codex model catalog (#2334 by @justram)
  • Added JSONL session export and import via /export <path.jsonl> and /import <path.jsonl> (#2356 by @hjanuschka)
  • Added a resizable sidebar to HTML share and export views (#2435 by @dmmulroy)

修复18

  • Tests for session-selector-rename and tree-selector are now keybinding-agnostic, resetting editor keybindings to defaults before each test so user keybindings.json cannot cause failures (#2360)
  • Fixed custom keybindings.json overrides to shadow conflicting default shortcuts globally, so bindings such as cursorUp: ["up", "ctrl+p"] no longer leave default actions like model cycling active (#2391)
  • Fixed concurrent edit and write mutations targeting the same file to run serially, preventing interleaved file writes from overwriting each other (#2327)
  • Fixed RPC mode to redirect unexpected stdout writes to stderr so JSONL responses remain parseable (#2388)
  • Fixed auto-retry with tool-using retry responses so session.prompt() waits for the full retry loop, including tool execution, before returning (#2440 by @pasky)
  • Fixed /model to refresh scoped model lists after models.json changes, avoiding stale selector contents (#2408 by @Perlence)
  • Fixed validateToolArguments() to fall back gracefully when AJV schema compilation is blocked in restricted runtimes such as Cloudflare Workers, allowing tool execution to proceed without schema validation (#2395)
  • Fixed CLI startup to suppress process warnings from leaking into terminal, print, and RPC output (#2404)
  • Fixed bash tool rendering to show elapsed time at the bottom of the tool block (#2406)
  • Fixed custom theme file watching to reload updated theme contents from disk instead of keeping stale cached theme data (#2417, #2003)
  • Fixed footer Git branch refreshes to run asynchronously so branch watcher updates do not block the UI (#2418)
  • Fixed invalid extension provider registrations to surface an extension error without preventing other providers from loading (#2431)
  • Fixed Windows bash execution hanging for commands that spawn detached descendants inheriting stdout/stderr handles, which caused agent-browser and similar commands to spin forever (#2389 by @mrexodia)
  • Fixed google-vertex API key resolution to ignore placeholder auth markers like <authenticated> and fall back to ADC instead of sending them as literal API keys (#2335)
  • Fixed desktop clipboard text copy to prefer native OS clipboard integration before shell fallbacks, improving reliability on macOS and Windows (#2347)
  • Fixed Bun Bedrock provider registration to survive provider resets and session reloads in compiled binaries (#2350 by @unexge)
  • Fixed OpenRouter reasoning requests to use the provider's nested reasoning payload, restoring thinking level support for OpenRouter models and custom compat settings (#2298 by @PriNova)
  • Fixed Bedrock application inference profiles to support prompt caching when AWS_BEDROCK_FORCE_CACHE=1 is set, covering profile ARNs that do not expose the underlying Claude model name (#2346 by @haoqixu)
0.60.0 2026-03-18 新功能 3破坏性变更 1新增 2修复 11 Fork existing sessions directly from the CLI with --fork <pa

新功能3

  • Fork existing sessions directly from the CLI with --fork <path|id>, which copies a source session into a new session in the current project. See README.md.
  • Extensions and SDK callers can reuse pi's built-in local bash backend via createLocalBashOperations() for user_bash interception and custom bash integrations. See docs/extensions.md#user_bash.
  • Startup no longer updates unpinned npm and git packages automatically. Use pi update explicitly, while interactive mode checks for updates in the background and notifies you when newer packages are available. See README.md.

破坏性变更1

  • Changed package startup behavior so installed unpinned packages are no longer checked or updated during startup. Use pi update to apply npm/git package updates, while interactive mode now checks for available package updates in the background and notifies you when updates are available (#1963)

新增2

  • Added --fork <path|id> CLI flag to fork an existing session file or partial session UUID directly into a new session (#2290)
  • Added createLocalBashOperations() export so extensions and SDK callers can wrap pi's built-in local bash backend for user_bash handling and other custom bash integrations (#2299)

修复11

  • Fixed active model selection to refresh immediately after dynamic provider registrations or updates change the available model set (#2291)
  • Fixed tmux xterm modifyOtherKeys matching for Backspace, Escape, and Space, and resolved raw \x08 backspace ambiguity by treating Windows Terminal sessions differently from legacy terminals (#2293)
  • Fixed Gemini 3 and Antigravity image tool results to stay inline as multimodal tool responses instead of being rerouted through separate follow-up messages (#2052)
  • Fixed bundled Bedrock Claude 4.6 model metadata to use the correct 200K context window instead of 1M (#2305)
  • Fixed /reload to reload keybindings from disk so changes in keybindings.json apply immediately (#2309)
  • Fixed lazy built-in provider registration so compiled Bun binaries can still load providers on first use without eagerly bundling provider SDKs (#2314)
  • Fixed built-in OAuth login flows to use aligned callback handling across Anthropic, Gemini CLI, Antigravity, and OpenAI Codex, and fixed OpenAI Codex login to complete immediately once the browser callback succeeds (#2316)
  • Fixed OpenAI-compatible z.ai network_error responses to trigger error handling and retries instead of being treated as successful assistant output (#2313)
  • Fixed print mode to merge piped stdin into the initial prompt when both stdin and an explicit prompt are provided (#2315)
  • Fixed OpenAI Responses replay in coding-agent to normalize oversized resumed tool call IDs before sending them back to OpenAI Codex and other Responses-compatible targets (#2328)
  • Fixed tmux extended-keys warning to stay hidden when the tmux server is unreachable, avoiding false startup warnings in sandboxed environments (#2311 by @kaffarell)
0.59.0 2026-03-17 新功能 4破坏性变更 1变更 1修复 5 Faster startup by lazy-loading @mariozechner/pi-ai provider

新功能4

  • Faster startup by lazy-loading @mariozechner/pi-ai provider SDKs on first use instead of import time (#2297)
  • Better provider retry behavior when providers return error messages as responses (#2264)
  • Better terminal integration via OSC 133 command-executed markers (#2242)
  • Better Git footer branch detection for repositories using reftable storage (#2300)

破坏性变更1

  • Changed custom tool system prompt behavior so extension and SDK tools are included in the default Available tools section only when they provide promptSnippet. Omitting promptSnippet now leaves the tool out of that section instead of falling back to description (#2285)

变更1

  • Lazy-load built-in @mariozechner/pi-ai provider modules and root provider wrappers so coding-agent startup no longer eagerly loads provider SDKs before first use (#2297)

修复5

  • Fixed session title handling in /tree, compaction, and branch summarization so empty title clears render correctly and session_info entries stay out of summaries (#2304 by @aliou)
  • Fixed footer branch detection for Git repositories using reftable storage so branch names still appear correctly in the footer (#2300)
  • Fixed rendered user messages to emit an OSC 133 command-executed marker after command output, improving terminal prompt integration (#2242)
  • Fixed provider retry handling to treat provider-returned error messages as retryable failures instead of successful responses (#2264)
  • Fixed Claude 4.6 context window overrides in bundled model metadata so coding-agent sees the intended model limits after generated catalogs are rebuilt (#2286)
0.58.4 2026-03-16 修复 1 Fixed steering messages to wait until the current assistant

修复1

  • Fixed steering messages to wait until the current assistant message's tool-call batch fully finishes instead of skipping pending tool calls.
0.58.3 2026-03-15 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.58.2 2026-03-15 新增 1修复 5 Improved settings, theme, thinking, and show-images selector

新增1

  • Improved settings, theme, thinking, and show-images selector layouts by using configurable select-list primary column sizing (#2154 by @markusylisiurunen)

修复5

  • Fixed fuzzy edit matching to normalize Unicode compatibility variants before comparison, reducing false "oldText not found" failures for text such as CJK and full-width characters (#2044)
  • Fixed /model <ref> exact matching and picker search to recognize canonical provider/model references when model IDs themselves contain /, such as LM Studio models like unsloth/qwen3.5-35b-a3b (#2174)
  • Fixed Anthropic OAuth manual login and token refresh by using the localhost callback URI for pasted redirect/code flows and omitting scope from refresh-token requests (#2169)
  • Fixed stale scrollback remaining after session switches by clearing the screen before wiping scrollback (#2155 by @Perlence)
  • Fixed extra blank lines after markdown block elements in rendered output (#2152 by @markusylisiurunen)
0.58.1 2026-03-14 新增 1修复 17 Added pi uninstall alias for pi install --uninstall convenie

新增1

  • Added pi uninstall alias for pi install --uninstall convenience

修复17

  • Fixed OpenAI Codex websocket protocol to include required headers and properly terminate SSE streams on connection close (#1961)
  • Fixed WSL clipboard image fallback to properly handle missing clipboard utilities and permission errors (#1722)
  • Fixed extension session_start hook firing before TUI was ready, causing UI operations in session_start handlers to fail (#2035)
  • Fixed Windows shell and path handling for package manager operations and autocomplete to properly handle drive letters and mixed path separators
  • Fixed Bedrock prompt caching being enabled for non-Claude models, causing API errors (#2053)
  • Fixed Qwen models via OpenAI-compatible providers by adding qwen-chat-template compat mode that uses Qwen's native chat template format (#2020)
  • Fixed Bedrock unsigned thinking replay to handle edge cases with empty or malformed thinking blocks (#2063)
  • Fixed headless clipboard fallback logging spurious errors in non-interactive environments (#2056)
  • Fixed models.json provider compat flags not being honored when loading custom model definitions (#2062)
  • Fixed xhigh reasoning effort detection for Claude Opus 4.6 to match by model ID instead of requiring explicit capability flag (#2040)
  • Fixed prompt cwd containing Windows backslashes breaking bash tool execution by normalizing to forward slashes (#2080)
  • Fixed editor paste to preserve literal content instead of normalizing newlines, preventing content corruption for text with embedded escape sequences (#2064)
  • Fixed skill discovery recursing past skill root directories when nested SKILL.md files exist (#2075)
  • Fixed tab completion to preserve ./ prefix when completing relative paths (#2087)
  • Fixed npm package installs and lookups being tied to the active repository Node version by adding npmCommand as an argv-style settings override for package manager operations (#2072)
  • Fixed ctx.ui.getEditorText() in the extension API returning paste markers (e.g., [paste #1 +24 lines]) instead of the actual pasted content (#2084)
  • Fixed startup crash when downloading fd/ripgrep on first run by using pipeline() instead of finished(readable.pipe(writable)) so stream errors from timeouts are caught properly, and increased the download timeout from 10s to 120s (#2066)
0.58.0 2026-03-14 新功能 4新增 2变更 2修复 16 Claude Opus 4.6, Sonnet 4.6, and related Bedrock models now

新功能4

  • Claude Opus 4.6, Sonnet 4.6, and related Bedrock models now use a 1M token context window (up from 200K) (#2135 by @mitsuhiko).
  • Extension tool calls now execute in parallel by default, with sequential tool_call preflight preserved for extension interception.
  • GOOGLE_CLOUD_API_KEY environment variable support for the google-vertex provider as an alternative to Application Default Credentials (#1976 by @gordonhwc).
  • Extensions can supply deterministic session IDs via newSession() (#2130 by @zhahaoyu).

新增2

  • Added GOOGLE_CLOUD_API_KEY environment variable support for the google-vertex provider as an alternative to Application Default Credentials (#1976 by @gordonhwc)
  • Added custom session ID support in newSession() for extensions that need deterministic session paths (#2130 by @zhahaoyu)

变更2

  • Changed extension tool interception to use agent-core beforeToolCall and afterToolCall hooks instead of wrapper-based interception. Tool calls now execute in parallel by default, extension tool_call preflight still runs sequentially, and final tool results are emitted in assistant source order.
  • Raised Claude Opus 4.6, Sonnet 4.6, and related Bedrock model context windows from 200K to 1M tokens (#2135 by @mitsuhiko)

修复16

  • Fixed tool_call extension handlers observing stale sessionManager state during multi-tool turns by draining queued agent events before each tool_call preflight. In parallel tool mode this guarantees state through the current assistant tool-calling message, but not sibling tool results from the same assistant message.
  • Fixed interactive input fields backed by the TUI Input component to scroll by visual column width for wide Unicode text (CJK, fullwidth characters), preventing rendered line overflow and TUI crashes in places like search and filter inputs (#1982)
  • Fixed shift+tab and other modified Tab bindings in tmux when extended-keys-format is left at the default xterm
  • Fixed EXIF orientation not being applied during image convert and resize, causing JPEG and WebP images from phone cameras to display rotated or mirrored (#2105 by @melihmucuk)
  • Fixed the default coding-agent system prompt to include only the current date in ISO format, not the current time, so prompt prefixes stay cacheable across reloads and resumed sessions (#2131)
  • Fixed retry regex to match server_error and internal_error error types from providers, improving automatic retry coverage (#2117 by @MadKangYu)
  • Fixed example extensions to support PI_CODING_AGENT_DIR environment variable for custom agent directory paths (#2009 by @smithbm2316)
  • Fixed tool result images not being sent in function_call_output items for OpenAI Responses API providers, causing image data to be silently dropped in tool results (#2104)
  • Fixed assistant content being sent as structured content blocks instead of plain strings in the openai-completions provider, causing errors with some OpenAI-compatible backends (#2008 by @geraldoaax)
  • Fixed error details in OpenAI Responses response.failed handler to include status code, error code, and message instead of a generic failure (#1956 by @drewburr)
  • Fixed GitHub Copilot device-code login polling to respect OAuth slow-down intervals, wait before the first token poll, and include a clearer clock-drift hint in WSL/VM environments when repeated slow-downs lead to timeout
  • Fixed usage statistics not being captured for OpenAI-compatible providers that return usage in choice.usage instead of the standard chunk.usage (e.g., Moonshot/Kimi) (#2017)
  • Fixed editor scroll indicator rendering crash in narrow terminal widths (#2103 by @haoqixu)
  • Fixed tab characters in editor and input paste not being normalized to spaces (#2027, #1975 by @haoqixu)
  • Fixed wordWrapLine overflow when wide characters (CJK, fullwidth) fall exactly at the wrap boundary (#2082 by @haoqixu)
  • Fixed paste markers not being treated as atomic segments in editor word wrapping and cursor navigation (#2111 by @haoqixu)
0.57.1 2026-03-07 新功能 3新增 3修复 7 Tree branch folding and segment-jump navigation in /tree, wi

新功能3

  • Tree branch folding and segment-jump navigation in /tree, with Ctrl+←/Ctrl+→ and Alt+←/Alt+→ shortcuts while ←/→ and Page Up/Page Down remain available for paging. See docs/tree.md and docs/keybindings.md.
  • session_directory extension event for customizing session directory paths before session manager creation. See docs/extensions.md.
  • Digit keybindings (0-9) in the TUI keybinding system, including modified combos like ctrl+1. See docs/keybindings.md.

新增3

  • Added /tree branch folding and segment-jump navigation with Ctrl+←/Ctrl+→ and Alt+←/Alt+→, while keeping ←/→ and Page Up/Page Down for paging (#1724 by @Perlence)
  • Added session_directory extension event that fires before session manager creation, allowing extensions to customize the session directory path based on cwd and other factors. CLI --session-dir flag takes precedence over extension-provided paths (#1730 by @hjanuschka).
  • Added digit keys (0-9) to the keybinding system, including Kitty CSI-u and xterm modifyOtherKeys support for bindings like ctrl+1 (#1905)

修复7

  • Fixed custom tool collapsed/expanded rendering in HTML exports. Custom tools that define different collapsed vs expanded displays now render correctly in exported HTML, with expandable sections when both states differ and direct display when only expanded exists (#1934 by @aliou)
  • Fixed tmux startup guidance and keyboard setup warnings for modified key handling, including Ghostty shift+enter=text:\n remap guidance and tmux extended-keys-format detection (#1872)
  • Fixed z.ai context overflow recovery so model_context_window_exceeded errors trigger auto-compaction instead of surfacing as unhandled stop reason failures (#1937)
  • Fixed autocomplete selection ignoring typed text: highlight now follows the first prefix match as the user types, and exact matches are always selected on Enter (#1931 by @aliou)
  • Fixed slash-command Tab completion to immediately open argument completions when available (#1481 by @barapa)
  • Fixed explicit pi -e <path> extensions losing command and tool conflicts to discovered extensions by giving CLI-loaded extensions higher precedence (#1896)
  • Fixed Windows external editor launch for Ctrl+G and ctx.ui.editor() so shell-based commands like EDITOR="code --wait" work correctly (#1925)
0.57.0 2026-03-07 新功能 3破坏性变更 1新增 2变更 1修复 3 Extensions can intercept and modify provider request payload

新功能3

  • Extensions can intercept and modify provider request payloads via before_provider_request. See docs/extensions.md#before_provider_request.
  • Extension UIs can use non-capturing overlays with explicit focus control via OverlayOptions.nonCapturing and OverlayHandle.focus() / unfocus() / isFocused(). See docs/extensions.md and ../tui/README.md.
  • RPC mode now uses strict LF-only JSONL framing for robust payload handling. See docs/rpc.md.

破坏性变更1

  • RPC mode now uses strict LF-delimited JSONL framing. Clients must split records on \n only instead of using generic line readers such as Node readline, which also split on Unicode separators inside JSON payloads (#1911)

新增2

  • Added before_provider_request extension hook so extensions can inspect or replace provider payloads before requests are sent, with an example in examples/extensions/provider-payload.ts
  • Added non-capturing overlay focus control for extension UIs via OverlayOptions.nonCapturing and OverlayHandle.focus() / unfocus() / isFocused() (#1916 by @nicobailon)

变更1

  • Overlay compositing in extension UIs now uses focus order so focused overlays render on top while preserving stack semantics for show/hide behavior (#1916 by @nicobailon)

修复3

  • Fixed RPC mode stdin/stdout framing to use strict LF-delimited JSONL instead of readline, so payloads containing U+2028 or U+2029 no longer corrupt command or event streams (#1911)
  • Fixed automatic overlay focus restoration in extension UIs to skip non-capturing overlays, and fixed overlay hide behavior to only reassign focus when the hidden overlay had focus (#1916 by @nicobailon)
  • Fixed pi config misclassifying ~/.agents/skills as project-scoped in non-git directories under $HOME, so toggling those skills no longer writes project overrides to .pi/settings.json (#1915)
0.56.3 2026-03-06 新功能 4新增 2修复 9 claude-sonnet-4-6 model available via the google-antigravity

新功能4

  • claude-sonnet-4-6 model available via the google-antigravity provider (#1859)
  • Custom editors can now define their own onEscape/onCtrlD handlers without being overwritten by app defaults, enabling vim-mode extensions (#1838)
  • Shift+Enter and Ctrl+Enter now work inside tmux via xterm modifyOtherKeys fallback (docs/tmux.md, #1872)
  • Auto-compaction is now resilient to persistent API errors (e.g. 529 overloaded) and no longer retriggers spuriously after compaction (#1834, #1860)

新增2

  • Added claude-sonnet-4-6 model for the google-antigravity provider (#1859).
  • Added tmux setup documentation for modified enter key support (#1872)

修复9

  • Fixed custom editors having their onEscape/onCtrlD handlers unconditionally overwritten by app-level defaults, making vim-style escape handling impossible (#1838)
  • Fixed auto-compaction retriggering on the first prompt after compaction due to stale pre-compaction assistant usage (#1860 by @joelhooks)
  • Fixed sessions never auto-compacting when hitting persistent API errors (e.g. 529 overloaded) by estimating context size from the last successful response (#1834)
  • Fixed compaction summarization requests exceeding context limits by truncating tool results to 2k chars (#1796)
  • Fixed /new leaving startup header content, including the changelog, visible after starting a fresh session (#1880)
  • Fixed misleading docs and example implying that returning { isError: true } from a tool's execute function marks the execution as failed; errors must be signaled by throwing (#1881)
  • Fixed model switches through non-reasoning models to preserve the saved default thinking level instead of persisting a capability-forced off clamp (#1864)
  • Fixed parallel pi processes failing with false "No API key found" errors due to immediate lockfile contention on auth.json and settings.json (#1871)
  • Fixed OpenAI Responses reasoning replay regression that broke multi-turn reasoning continuity (#1878)
0.56.2 2026-03-05 新功能 3新增 3变更 1修复 10 GPT-5.4 support across openai, openai-codex, azure-openai-re

新功能3

  • GPT-5.4 support across openai, openai-codex, azure-openai-responses, and opencode, with gpt-5.4 now the default for openai and openai-codex (README.md, docs/providers.md).
  • treeFilterMode setting to choose the default /tree filter mode (default, no-tools, user-only, labeled-only, all) (docs/settings.md, #1852 by @lajarre).
  • Mistral native conversations integration with SDK-backed provider behavior, preserving Mistral-specific thinking and replay semantics (README.md, docs/providers.md, #1716).

新增3

  • Added gpt-5.4 model availability for openai, openai-codex, azure-openai-responses, and opencode providers.
  • Added gpt-5.3-codex fallback model availability for github-copilot until upstream model catalogs include it (#1853).
  • Added treeFilterMode setting to choose the default /tree filter mode (default, no-tools, user-only, labeled-only, all) (#1852 by @lajarre).

变更1

  • Updated the default models for the openai and openai-codex providers to gpt-5.4.

修复10

  • Fixed GPT-5.3 Codex follow-up turns dropping OpenAI Responses assistant phase metadata by preserving replayable signatures in session history and forwarding phase back to the Responses API (#1819).
  • Fixed OpenAI Responses replay to omit empty thinking blocks, avoiding invalid no-op reasoning items in follow-up turns.
  • Updated Mistral integration to use the native SDK-backed provider and conversations API, including coding-agent model/provider wiring and Mistral setup documentation (#1716).
  • Fixed Antigravity reliability: endpoint cascade on 403/404, added autopush sandbox fallback, removed extra fingerprint headers (#1830).
  • Fixed @mariozechner/pi-ai/oauth extension imports in published installs by resolving the subpath directly from built dist files instead of package-root wrapper shims (#1856).
  • Fixed Gemini 3 multi-turn tool use losing structured context by using skip_thought_signature_validator sentinel for unsigned function calls instead of text fallback (#1829).
  • Fixed model selector filter not accepting typed characters in VS Code 1.110+ due to missing Kitty CSI-u printable decoding in the Input component (#1857)
  • Fixed editor/footer visibility drift during terminal resize by forcing full redraws when terminal width or height changes (#1844 by @ghoulr).
  • Fixed footer width truncation for wide Unicode text (session name, model, provider) to prevent TUI crashes from rendered lines exceeding terminal width (#1833).
  • Fixed Windows write preview background artifacts by normalizing CRLF content (\r\n) to LF for display rendering in tool output previews (#1854).
0.56.1 2026-03-05 修复 2 Fixed extension alias fallback resolution to use ESM-aware r

修复2

  • Fixed extension alias fallback resolution to use ESM-aware resolution for jiti aliases in global installs (#1821 by @Perlence)
  • Fixed markdown blockquote rendering to isolate blockquote styling from default text style, preventing style leakage.
0.56.0 2026-03-04 新功能 3破坏性变更 2新增 3变更 1修复 18 Added OpenCode Go provider support with opencode-go model de

新功能3

  • Added OpenCode Go provider support with opencode-go model defaults and OPENCODE_API_KEY environment variable support (docs/providers.md, #1757).
  • Added branchSummary.skipPrompt setting to skip branch summarization prompts during tree navigation (docs/settings.md, #1792).
  • Added gemini-3.1-flash-lite-preview fallback model availability for Google provider catalogs when upstream model metadata lags (README.md, #1785).

破坏性变更2

  • Changed scoped model thinking semantics. Scoped entries without an explicit :<thinking> suffix now inherit the current session thinking level when selected, instead of applying a startup-captured default.
  • Moved Node OAuth runtime exports off the top-level @mariozechner/pi-ai entry. OAuth login and refresh must be imported from @mariozechner/pi-ai/oauth (#1814).

新增3

  • Added branchSummary.skipPrompt setting to skip the summary prompt when navigating branches (#1792).
  • Added OpenCode Go provider support with opencode-go model defaults and OPENCODE_API_KEY environment variable support (#1757).
  • Added gemini-3.1-flash-lite-preview fallback model availability in provider catalogs when upstream catalogs lag (#1785).

变更1

  • Updated Antigravity Gemini 3.1 model metadata and request headers to match upstream behavior.

修复18

  • Fixed IME hardware cursor positioning in the custom extension editor (ctx.ui.editor() / extension editor dialog) by propagating focus to the internal Editor, preventing the terminal cursor from getting stuck at the bottom-right during composition.
  • Added OSC 133 semantic zone markers around rendered user messages to support terminal navigation between prompts in iTerm2, WezTerm, Kitty, Ghostty, and other compatible terminals (#1805).
  • Fixed markdown blockquotes dropping nested list content in the TUI renderer (#1787).
  • Fixed TUI width handling for regional indicator symbols to prevent wrap drift and stale characters during streaming (#1783).
  • Fixed Kitty CSI-u handling to ignore unsupported modifiers so modifier-only events do not insert printable characters (#1807).
  • Fixed single-line paste handling to insert text atomically and avoid repeated @ autocomplete scans on large pastes (#1812).
  • Fixed extension loading with the new @mariozechner/pi-ai/oauth export path by aliasing the oauth subpath in the extension loader and development path mapping (#1814).
  • Fixed browser-safe provider loading regressions by preloading the Bedrock provider module in compiled Bun binaries and rebuilding binaries against fresh workspace dependencies (#1814).
  • Fixed GNU screen terminal detection by downgrading theme output to 256-color mode for screen* TERM values (#1809).
  • Fixed branch summarization queue handling so messages typed while summaries are generated are processed correctly (#1803).
  • Fixed compaction summary requests to avoid reasoning output for non-reasoning models (#1793).
  • Fixed overflow auto-compaction cascades so a single overflow does not trigger repeated compaction loops.
  • Fixed models.json to allow provider-scoped custom model ids and model-level baseUrl overrides (#1759, #1777).
  • Fixed session selector display sanitization by stripping control characters from session display text (#1747).
  • Fixed Groq Qwen3 reasoning effort mapping for OpenAI-compatible models (#1745).
  • Fixed Bedrock AWS_PROFILE region resolution by honoring profile region values (#1800).
  • Fixed Gemini 3.1 thinking-level detection for google and google-vertex providers (#1785).
  • Fixed browser bundling compatibility for @mariozechner/pi-ai by removing Node-only side effects from default browser import paths (#1814).
0.55.4 2026-03-02 新功能 3新增 2修复 4 Runtime tool registration now applies immediately in active

新功能3

  • Runtime tool registration now applies immediately in active sessions. Tools registered via pi.registerTool() after startup are available to pi.getAllTools() and the LLM without /reload (docs/extensions.md, examples/extensions/dynamic-tools.ts, #1720).
  • Tool definitions can customize the default system prompt with promptSnippet (Available tools) and promptGuidelines (Guidelines) while the tool is active (docs/extensions.md, #1720).
  • Custom tool renderers can suppress transcript output without leaving extra spacing or empty transcript footprint in interactive rendering (docs/extensions.md, #1719).

新增2

  • Added optional promptSnippet to ToolDefinition for one-line entries in the default system prompt's Available tools section. Active extension tools appear there when registered and active (#1237 by @semtexzv).
  • Added optional promptGuidelines to ToolDefinition so active tools can append tool-specific bullets to the default system prompt Guidelines section (#1720).

修复4

  • Fixed pi.registerTool() dynamic registration after session initialization. Tools registered in session_start and later handlers now refresh immediately, become active, and are visible to the LLM without /reload (#1720)
  • Fixed session message persistence ordering by serializing AgentSession event processing, preventing toolResult entries from being written before their corresponding assistant tool-call messages when extension handlers are asynchronous (#1717)
  • Fixed spacing artifacts when custom tool renderers intentionally suppress per-call transcript output, including extra blank rows in interactive streaming and non-zero transcript footprint for empty custom renders (#1719 by @alasano)
  • Fixed session.prompt() returning before retry completion by creating the retry promise synchronously at agent_end dispatch, which closes a race when earlier queued event handlers are async (#1726 by @pasky)
2026-0231 个版本
0.55.3 2026-02-27 修复 1 Changed the default image paste keybinding on Windows to alt

修复1

  • Changed the default image paste keybinding on Windows to alt+v to avoid ctrl+v conflicts with terminal paste behavior (#1682 by @mrexodia).
0.55.2 2026-02-27 新功能 2新增 1修复 7 Extensions can dynamically remove custom providers via pi.un

新功能2

  • Extensions can dynamically remove custom providers via pi.unregisterProvider(name), restoring any built-in models that were overridden, without requiring /reload (docs).
  • pi.registerProvider() now takes effect immediately when called outside the initial extension load phase (e.g. from a command handler), removing the need for /reload after late registrations.

新增1

  • pi.unregisterProvider(name) removes a dynamically registered provider and its models from the registry without requiring /reload. Built-in models that were overridden by the provider are restored (#1669 by @aliou).

修复7

  • pi.registerProvider() now takes effect immediately when called after the initial extension load phase (e.g. from a command handler). Previously the registration sat in a pending queue that was never flushed until the next /reload (#1669 by @aliou).
  • Fixed duplicate session headers when forking from a point before any assistant message. createBranchedSession now defers file creation to _persist() when the branched path has no assistant message, matching the newSession() contract (#1672 by @w-winter).
  • Fixed SIGINT being delivered to pi while the process is suspended (e.g. via ctrl+z), which could corrupt terminal state on resume (#1668 by @aliou).
  • Fixed Z.ai thinking control using wrong parameter name, causing thinking to always be enabled and wasting tokens/latency (#1674 by @okuyam2y)
  • Fixed redacted_thinking blocks being silently dropped during Anthropic streaming, and related issues with interleaved-thinking beta headers and temperature being sent alongside extended thinking (#1665 by @tctev)
  • Fixed (external, cli) user-agent flag causing 401 errors on Anthropic setup-token endpoint (#1677 by @LazerLance777)
  • Fixed crash when OpenAI-compatible provider returns a chunk with no choices array (#1671)
0.55.1 2026-02-26 新功能 2修复 7 Added offline startup mode via --offline (or PI_OFFLINE) to

新功能2

  • Added offline startup mode via --offline (or PI_OFFLINE) to disable startup network operations, with startup network timeouts to avoid hangs in restricted or offline environments.
  • Added gemini-3.1-pro-preview model support to the google-gemini-cli provider (#1599 by @audichuang).

修复7

  • Fixed offline startup hangs by adding offline startup behavior and network timeouts during managed tool setup (#1631 by @mcollina)
  • Fixed Windows VT input initialization in ESM by loading koffi via createRequire, avoiding runtime and bundling issues in end-user environments (#1627 by @kaste)
  • Fixed managed fd/rg bootstrap on Windows in Git Bash by using extract-zip for .zip archives, searching extracted layouts more robustly, and isolating extraction temp directories to avoid concurrent download races (#1348)
  • Fixed extension loading on Windows when resolving @sinclair/typebox aliases so subpath imports like @sinclair/typebox/compiler resolve correctly.
  • Fixed adaptive thinking for Claude Sonnet 4.6 in Anthropic and Bedrock providers, and clamped unsupported xhigh effort values to supported levels (#1548 by @tctev)
  • Fixed Vertex ADC credential detection race by avoiding caching a false negative during async import initialization (#1550 by @jeremiahgaylord-web)
  • Fixed subagent extension example to resolve user agents from the configured agent directory instead of hardcoded paths (#1559 by @tianshuwang)
0.55.0 2026-02-24 破坏性变更 2 Resource precedence for extensions, skills, prompts, themes,

破坏性变更2

  • Resource precedence for extensions, skills, prompts, themes, and slash-command name collisions is now project-first (cwd/.pi) before user-global (~/.pi/agent). If you relied on global resources overriding project resources with the same names, rename or reorder your resources.
  • Extension registration conflicts no longer unload the entire later extension. All extensions stay loaded, and conflicting command/tool/flag names are resolved by first registration in load order.
0.54.2 2026-02-23 修复 3 Fixed .pi folder being created unnecessarily when only readi

修复3

  • Fixed .pi folder being created unnecessarily when only reading settings. The folder is now only created when writing project-specific settings.
  • Fixed extension-driven runtime theme changes to persist in settings so /settings reflects the active currentTheme after ctx.ui.setTheme(...) (#1483 by @ferologics)
  • Fixed interactive mode freezes during large streaming write tool calls by using incremental syntax highlighting while partial arguments stream, with a final full re-highlight after tool-call arguments complete.
0.54.1 2026-02-22 修复 1 Externalized koffi from bun binary builds, reducing archive

修复1

  • Externalized koffi from bun binary builds, reducing archive sizes by ~15MB per platform (e.g. darwin-arm64: 43MB -> 28MB). Koffi's Windows-only .node file is now shipped alongside the Windows binary only.
0.54.0 2026-02-19 新增 1 Added default skill auto-discovery for .agents/skills locati

新增1

  • Added default skill auto-discovery for .agents/skills locations. Pi now discovers project skills from .agents/skills in cwd and ancestor directories (up to git repo root, or filesystem root when not in a repo), and global skills from ~/.agents/skills, in addition to existing .pi skill paths.
0.53.1 2026-02-19 变更 2 Added Gemini 3.1 model catalog entries for all built-in prov

变更2

  • Added Gemini 3.1 model catalog entries for all built-in providers that currently expose it: google, google-vertex, opencode, openrouter, and vercel-ai-gateway.
  • Added Claude Opus 4.6 Thinking to the google-antigravity model catalog.
0.53.0 2026-02-17 破坏性变更 2新增 3变更 1修复 3 SettingsManager persistence semantics changed for SDK consum

破坏性变更2

  • SettingsManager persistence semantics changed for SDK consumers. Setters now update in-memory state immediately and queue disk writes. Code that requires durable on-disk settings must call await settingsManager.flush().
  • AuthStorage constructor is no longer public. Use static factories (AuthStorage.create(...), AuthStorage.fromStorage(...), AuthStorage.inMemory(...)). This breaks code that used new AuthStorage(...) directly.

新增3

  • Added SettingsManager.drainErrors() for caller-controlled settings I/O error handling without manager-side console output.
  • Added auth storage backends (FileAuthStorageBackend, InMemoryAuthStorageBackend) and AuthStorage.fromStorage(...) for storage-first auth persistence wiring.
  • Added Anthropic claude-sonnet-4-6 model fallback entry to generated model definitions.

变更1

  • SettingsManager now uses scoped storage abstraction with per-scope locked read/merge/write persistence for global and project settings.

修复3

  • Fixed project settings persistence to preserve unrelated external edits via merge-on-write, while still applying in-memory changes for modified keys.
  • Fixed auth credential persistence to preserve unrelated external edits to auth.json via locked read/merge/write updates.
  • Fixed auth load/persist error surfacing by buffering errors and exposing them via AuthStorage.drainErrors().
0.52.12 2026-02-13 新增 1变更 2 Added transport setting ("sse", "websocket", "auto") to /set

新增1

  • Added transport setting ("sse", "websocket", "auto") to /settings and settings.json for providers that support multiple transports (currently openai-codex via OpenAI Codex Responses).

变更2

  • Interactive mode now applies transport changes immediately to the active agent session.
  • Settings migration now maps legacy websockets: boolean to the new transport setting.
0.52.11 2026-02-13 新增 1 Added MiniMax M2.5 model entries for minimax, minimax-cn, op

新增1

  • Added MiniMax M2.5 model entries for minimax, minimax-cn, openrouter, and vercel-ai-gateway providers, plus minimax-m2.5-free for opencode.
0.52.10 2026-02-12 新功能 6破坏性变更 2新增 3变更 1修复 8 Extension terminal input interception via terminal_input, al

新功能6

  • Extension terminal input interception via terminal_input, allowing extensions to consume or transform raw input before normal TUI handling. See docs/extensions.md.
  • Expanded CLI model selection: --model now supports provider/id, fuzzy matching, and :<thinking> suffixes. See README.md and docs/models.md.
  • Safer package source handling with stricter git source parsing and improved local path normalization. See docs/packages.md.
  • New built-in model definition gpt-5.3-codex-spark for OpenAI and OpenAI Codex providers.
  • Improved OpenAI stream robustness for malformed trailing tool-call JSON in partial chunks.
  • Added built-in GLM-5 model support via z.ai and OpenRouter provider catalogs.

破坏性变更2

  • ContextUsage.tokens and ContextUsage.percent are now number | null. After compaction, context token count is unknown until the next LLM response, so these fields return null. Extensions that read ContextUsage must handle the null case. Removed usageTokens, trailingTokens, and lastUsageIndex fields from ContextUsage (implementation details that should not have been public) (#1382 by @ferologics)
  • Git source parsing is now strict without git: prefix: only protocol URLs are treated as git (https://, http://, ssh://, git://). Shorthand sources like github.com/org/repo and git@github.com:org/repo now require the git: prefix. (#1426)

新增3

  • Added extension event forwarding for message and tool execution lifecycles (message_start, message_update, message_end, tool_execution_start, tool_execution_update, tool_execution_end) (#1375 by @sumeet)
  • Added terminal_input extension event to intercept, consume, or transform raw terminal input before normal TUI handling.
  • Added gpt-5.3-codex-spark model definition for OpenAI and OpenAI Codex providers (research preview).

变更1

  • Routed GitHub Copilot Claude 4.x models through Anthropic Messages API, with updated Copilot header handling for Claude model requests.

修复8

  • Fixed context usage percentage in footer showing stale pre-compaction values. After compaction the footer now shows ?/200k until the next LLM response provides accurate usage (#1382 by @ferologics)
  • Fixed _checkCompaction() using the first compaction entry instead of the latest, which could cause incorrect overflow detection with multiple compactions (#1382 by @ferologics)
  • --model now works without --provider, supports provider/id syntax, fuzzy matching, and :<thinking> suffix (e.g., --model sonnet:high, --model openai/gpt-4o) (#1350 by @mitsuhiko)
  • Fixed local package path normalization for extension sources while tightening git source parsing rules (#1426)
  • Fixed extension terminal input listeners not being cleared during session resets, which could leave stale handlers active.
  • Fixed Termux bootstrap package name for fd installation (#1433)
  • Fixed @ file autocomplete fuzzy matching to prioritize path-prefix and segment matches for nested paths (#1423)
  • Fixed OpenAI streaming tool-call parsing to tolerate malformed trailing JSON in partial chunks (#1424)
0.52.9 2026-02-08 新功能 4新增 4修复 10 Extensions can trigger a full runtime reload via ctx.reload(

新功能4

  • Extensions can trigger a full runtime reload via ctx.reload(), useful for hot-reloading configuration or restarting the agent. See docs/extensions.md and the C1 example (#1371)
  • Short CLI disable aliases: -ne (--no-extensions), -ns (--no-skills), and -np (--no-prompt-templates) for faster interactive usage and scripting.
  • /export HTML now includes collapsible tool input schemas (parameter names, types, and descriptions), improving session review and sharing workflows (#1416 by @marchellodev).
  • pi.getAllTools() now exposes tool parameters in addition to name and description, enabling richer extension integrations (#1416 by @marchellodev).

新增4

  • Added ctx.reload() to the extension API for programmatic runtime reload (#1371)
  • Added short aliases for disable flags: -ne for --no-extensions, -ns for --no-skills, -np for --no-prompt-templates
  • /export HTML now includes tool input schema (parameter names, types, descriptions) in a collapsible section under each tool (#1416 by @marchellodev)
  • pi.getAllTools() now returns tool parameters in addition to name and description (#1416 by @marchellodev)

修复10

  • Fixed extension source parsing so dot-prefixed local paths (for example .pi/extensions/foo.ts) are treated as local paths instead of git URLs
  • Fixed fd/rg download failing on Windows due to unzip not being available; now uses tar for both .tar.gz and .zip extraction, with proper error reporting (#1348)
  • Fixed RPC mode documentation incorrectly stating ctx.hasUI is false; it is true because dialog and fire-and-forget UI methods work via the RPC sub-protocol. Also documented missing unsupported/degraded methods (pasteToEditor, getAllThemes, getTheme, setTheme) (#1411 by @aliou)
  • Fixed rg not available in bash tool by downloading it at startup alongside fd (#1348)
  • Fixed custom-compaction example to use ModelRegistry (#1387)
  • Google providers now support full JSON Schema in tool declarations (anyOf, oneOf, const, etc.) (#1398 by @jarib)
  • Reverted incorrect Antigravity model change: claude-opus-4-6-thinking back to claude-opus-4-5-thinking (model does not exist on Antigravity endpoint)
  • Updated the Antigravity system instruction to a more compact version for Google Gemini CLI compatibility
  • Corrected opencode context windows for Claude Sonnet 4 and 4.5 (#1383)
  • Fixed subagent example unknown-agent errors to include available agent names (#1414 by @dnouri)
0.52.8 2026-02-07 新功能 4新增 4变更 1修复 4 Emacs-style kill ring (ctrl+k/ctrl+y/alt+y) and undo (ctrl+z

新功能4

  • Emacs-style kill ring (ctrl+k/ctrl+y/alt+y) and undo (ctrl+z) in the editor input (#1373 by @Perlence)
  • OpenRouter auto model alias (openrouter:auto) for automatic model routing (#1361 by @yogasanas)
  • Extensions can programmatically paste content into the editor via pasteToEditor in the extension UI context. See docs/extensions.md (#1351 by @kaofelix)
  • pi <package> --help and invalid subcommands now show helpful output instead of failing silently (#1347 by @ferologics)

新增4

  • Added pasteToEditor to extension UI context for programmatic editor paste (#1351 by @kaofelix)
  • Added package subcommand help and friendly error messages for invalid commands (#1347 by @ferologics)
  • Added OpenRouter auto model alias for automatic model routing (#1361 by @yogasanas)
  • Added kill ring (ctrl+k/ctrl+y/alt+y) and undo (ctrl+z) support to the editor input (#1373 by @Perlence)

变更1

修复4

  • Fixed temporary git package caches (-e <git-url>) to refresh on cache hits for unpinned sources, including detached/no-upstream checkouts
  • Fixed aborting retries when an extension customizes the editor (#1364 by @Perlence)
  • Fixed autocomplete not propagating to custom editors created by extensions (#1372 by @Perlence)
  • Fixed extension shutdown to use clean TUI shutdown path, preventing orphaned processes
0.52.7 2026-02-06 新功能 3破坏性变更 1新增 2修复 8 Per-model overrides in models.json via modelOverrides, allow

新功能3

  • Per-model overrides in models.json via modelOverrides, allowing customization of built-in provider models without replacing provider model lists. See docs/models.md#per-model-overrides.
  • models.json provider models now merge with built-in models by id, so custom models can be added or replace matching built-ins without full provider replacement. See docs/models.md#overriding-built-in-providers.
  • Bedrock proxy support for unauthenticated endpoints via AWS_BEDROCK_SKIP_AUTH and AWS_BEDROCK_FORCE_HTTP1. See docs/providers.md.

破坏性变更1

  • Changed models.json provider models behavior from full replacement to merge-by-id with built-in models. Built-in models are now kept by default, and custom models upsert by id.

新增2

  • Added modelOverrides in models.json to customize individual built-in models per provider without full provider replacement (#1332 by @charles-cooper)
  • Added AWS_BEDROCK_SKIP_AUTH and AWS_BEDROCK_FORCE_HTTP1 environment variables for connecting to unauthenticated Bedrock proxies (#1320 by @virtuald)

修复8

  • Fixed extra spacing between thinking-only assistant content and subsequent tool execution blocks when assistant messages contain no text
  • Fixed queued steering/follow-up/custom messages remaining stuck after threshold auto-compaction by resuming the agent loop when Agent-level queues still contain pending messages (#1312 by @ferologics)
  • Fixed tool_result extension handlers to chain result patches across handlers instead of last-handler-wins behavior (#1280)
  • Fixed compromised auth lock files being handled gracefully instead of crashing auth storage initialization (#1322)
  • Fixed Bedrock adaptive thinking handling for Claude Opus 4.6 with interleaved thinking beta responses (#1323 by @markusylisiurunen)
  • Fixed OpenAI Responses API requests to use store: false by default to avoid server-side history logging (#1308)
  • Fixed interactive mode startup by initializing autocomplete after resources are loaded (#1328)
  • Fixed modelOverrides merge behavior for nested objects and documented usage details (#1062)
0.52.6 2026-02-05 破坏性变更 1修复 2 Removed /exit command handling. Use /quit to exit (#1303)

破坏性变更1

  • Removed /exit command handling. Use /quit to exit (#1303)

修复2

  • Fixed /quit being shadowed by fuzzy slash command autocomplete matches from skills by adding /quit to built-in command autocomplete (#1303)
  • Fixed local package source parsing and settings normalization regression that misclassified relative paths as git URLs and prevented globally installed local packages from loading after restart (#1304)
0.52.5 2026-02-05 修复 1 Fixed thinking level capability detection so Anthropic Opus

修复1

  • Fixed thinking level capability detection so Anthropic Opus 4.6 models expose xhigh in selectors and cycling
0.52.4 2026-02-05 修复 1 Fixed extensions setting not respecting package.json pi.exte

修复1

  • Fixed extensions setting not respecting package.json pi.extensions manifest when directory is specified directly (#1302 by @hjanuschka)
0.52.3 2026-02-05 修复 5 Fixed git package parsing fallback for unknown hosts so ente

修复5

  • Fixed git package parsing fallback for unknown hosts so enterprise git sources like git:github.tools.sap/org/repo are treated as git packages instead of local paths
  • Fixed git package @ref parsing for shorthand, HTTPS, and SSH source formats, including branch refs with slashes
  • Fixed Bedrock default model ID from us.anthropic.claude-opus-4-6-v1:0 to us.anthropic.claude-opus-4-6-v1
  • Fixed Bedrock Opus 4.6 model metadata (IDs, cache pricing) and added missing EU profile
  • Fixed Claude Opus 4.6 context window metadata to 200000 for Anthropic and OpenCode providers
0.52.2 2026-02-05 变更 5 Updated default model for anthropic provider to claude-opus-

变更5

  • Updated default model for anthropic provider to claude-opus-4-6
  • Updated default model for openai-codex provider to gpt-5.3-codex
  • Updated default model for amazon-bedrock provider to us.anthropic.claude-opus-4-6-v1:0
  • Updated default model for vercel-ai-gateway provider to anthropic/claude-opus-4-6
  • Updated default model for opencode provider to claude-opus-4-6
0.52.1 2026-02-05 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.52.0 2026-02-05 新功能 5新增 6修复 6 Claude Opus 4.6 model support.

新功能5

  • Claude Opus 4.6 model support.
  • GPT-5.3 Codex model support (OpenAI Codex provider only).
  • SSH URL support for git packages. See docs/packages.md.
  • auth.json API keys now support shell command resolution (!command) and environment variable lookup. See docs/providers.md.
  • Model selectors now display the selected model name.

新增6

  • API keys in auth.json now support shell command resolution (!command) and environment variable lookup, matching the behavior in models.json
  • Added minimal-mode.ts example extension demonstrating how to override built-in tool rendering for a minimal display mode
  • Added Claude Opus 4.6 model to the model catalog
  • Added GPT-5.3 Codex model to the model catalog (OpenAI Codex provider only)
  • Added SSH URL support for git packages (#1287 by @markusn)
  • Model selectors now display the selected model name (#1275 by @haoqixu)

修复6

  • Fixed HTML export losing indentation in ANSI-rendered tool output (e.g. JSON code blocks in custom tool results) (#1269 by @aliou)
  • Fixed images being silently dropped when prompt() is called with both images and streamingBehavior during streaming. steer(), followUp(), and the corresponding RPC commands now accept optional images. (#1271 by @aliou)
  • CLI --help, --version, --list-models, and --export now exit even if extensions keep the event loop alive (#1285 by @ferologics)
  • Fixed crash when models send malformed tool arguments (objects instead of strings) (#1259)
  • Fixed custom message expand state not being respected (#1258 by @Gurpartap)
  • Fixed skill loader to respect .gitignore, .ignore, and .fdignore when scanning directories
0.51.6 2026-02-04 新功能 1新增 1变更 1修复 8 Configurable resume keybinding action for opening the sessio

新功能1

  • Configurable resume keybinding action for opening the session resume selector. See docs/keybindings.md. (#1249 by @juanibiapina)

新增1

  • Added resume as a configurable keybinding action, allowing users to bind a key to open the session resume selector (like newSession, tree, and fork) (#1249 by @juanibiapina)

变更1

  • Slash command menu now triggers on the first line even when other lines have content, allowing commands to be prepended to existing text (#1227 by @aliou)

修复8

  • Ignored unknown skill frontmatter fields when loading skills
  • Fixed /reload not picking up changes in global settings.json (#1241)
  • Fixed forked sessions to persist the user message after forking
  • Fixed forked sessions to write to new session files instead of the parent (#1242)
  • Fixed local package removal to normalize paths before comparison (#1243)
  • Fixed OpenAI Codex Responses provider to respect configured baseUrl (#1244)
  • Fixed /settings crashing in narrow terminals by handling small widths in the settings list (#1246 by @haoqixu)
  • Fixed Unix bash detection to fall back to PATH lookup when /bin/bash is unavailable, including Termux setups (#1230 by @VaclavSynacek)
0.51.5 2026-02-04 变更 1修复 1 Changed Bedrock model generation to drop legacy workarounds

变更1

  • Changed Bedrock model generation to drop legacy workarounds now handled upstream (#1239 by @unexge)

修复1

  • Fixed Windows package installs regression by using shell execution instead of .cmd resolution (#1220)
0.51.4 2026-02-03 新功能 1变更 1修复 1 Share URLs now default to pi.dev, graciously donated by exe.

新功能1

  • Share URLs now default to pi.dev, graciously donated by exe.dev.

变更1

  • Share URLs now use pi.dev by default while pi.dev and buildwithpi.ai continue to work.

修复1

  • Fixed input scrolling to avoid splitting emoji sequences (#1228 by @haoqixu)
0.51.3 2026-02-03 新功能 2破坏性变更 1新增 3修复 3 Command discovery for extensions via ExtensionAPI.getCommand

新功能2

  • Command discovery for extensions via ExtensionAPI.getCommands(), with commands.ts example for invocation patterns. See docs/extensions.md#pigetcommands and examples/extensions/commands.ts.
  • Local path support for pi install and pi remove, with relative path resolution against the settings file. See docs/packages.md#local-paths.

破坏性变更1

  • RPC get_commands response and SlashCommandSource type: renamed "template" to "prompt" for consistency with the rest of the codebase

新增3

  • Added ExtensionAPI.getCommands() to let extensions list available slash commands (extensions, prompt templates, skills) for invocation via prompt (#1210 by @w-winter)
  • Added commands.ts example extension and exported SlashCommandInfo types for command discovery integrations (#1210 by @w-winter)
  • Added local path support for pi install and pi remove with relative paths stored against the target settings file (#1216)

修复3

  • Fixed default thinking level persistence so settings-derived defaults are saved and restored correctly
  • Fixed Windows package installs by resolving npm.cmd when npm is not directly executable (#1220)
  • Fixed xhigh thinking level support check to accept gpt-5.2 model IDs (#1209)
0.51.2 2026-02-03 新功能 1新增 2修复 7 Extension tool output expansion controls via ExtensionUICont

新功能1

  • Extension tool output expansion controls via ExtensionUIContext getToolsExpanded and setToolsExpanded. See docs/extensions.md and docs/rpc.md.

新增2

  • Added ExtensionUIContext getToolsExpanded and setToolsExpanded for controlling tool output expansion (#1199 by @academo)
  • Added install method detection to show package manager specific update instructions (#1203 by @Itsnotaka)

修复7

  • Fixed Kitty key release events leaking to parent shell over slow SSH connections by draining stdin for up to 1s on exit (#1204)
  • Fixed legacy newline handling in the editor to preserve previous newline behavior
  • Fixed @ autocomplete to include hidden paths
  • Fixed submit fallback to honor configured keybindings
  • Fixed extension commands conflicting with built-in commands by skipping them (#1196 by @haoqixu)
  • Fixed @-prefixed tool paths failing to resolve by stripping the prefix (#1206)
  • Fixed install method detection to avoid stale cached results
0.51.1 2026-02-02 新功能 2修复 3 Extension API switchSession

新功能2

  • Extension API switchSession: Extensions can now programmatically switch sessions via ctx.switchSession(sessionPath). See docs/extensions.md. (#1187)
  • Clear on shrink setting: New terminal.clearOnShrink setting keeps the editor and footer pinned to the bottom of the terminal when content shrinks. May cause some flicker due to redraws. Disabled by default. Enable via /settings or PI_CLEAR_ON_SHRINK=1 env var.

修复3

  • Fixed scoped models not finding valid credentials after logout (#1194 by @terrorobe)
  • Fixed Ctrl+D exit closing the parent SSH session due to stdin buffer race condition (#1185)
  • Fixed emoji cursor positioning in editor input (#1183 by @haoqixu)
0.51.0 2026-02-01 破坏性变更 1新功能 7新增 10修复 8 Extension tool signature change

破坏性变更1

  • Extension tool signature change: ToolDefinition.execute now uses (toolCallId, params, signal, onUpdate, ctx) parameter order to match AgentTool.execute. Previously it was (toolCallId, params, onUpdate, ctx, signal). This makes wrapping built-in tools trivial since the first four parameters now align. Update your extensions by swapping the signal and onUpdate parameters:
    // Before
async execute(toolCallId, params, onUpdate, ctx, signal) { ... }
// After
async execute(toolCallId, params, signal, onUpdate, ctx) { ... }

新功能7

  • Android/Termux support: Pi now runs on Android via Termux. Install with:
    pkg install nodejs termux-api git
npm install -g @mariozechner/pi-coding-agent
mkdir -p ~/.pi/agent
echo "You are running on Android in Termux." > ~/.pi/agent/AGENTS.md

    Clipboard operations fall back gracefully when termux-api is unavailable. (#1164)

  • Bash spawn hook: Extensions can now intercept and modify bash commands before execution via pi.setBashSpawnHook(). Adjust the command string, working directory, or environment variables. See docs/extensions.md. (#1160 by @mitsuhiko)
  • Linux ARM64 musl support: Pi now runs on Alpine Linux ARM64 (linux-arm64-musl) via updated clipboard dependency.
  • Nix/Guix support: PI_PACKAGE_DIR environment variable overrides the package path for content-addressed package managers where store paths tokenize poorly. See README.md#environment-variables. (#1153 by @odysseus0)
  • Named session filter: /resume picker now supports filtering to show only named sessions via Ctrl+N. Configurable via toggleSessionNamedFilter keybinding. See docs/keybindings.md. (#1128 by @w-winter)
  • Typed tool call events: Extension developers can narrow ToolCallEvent types using isToolCallEventType() for better TypeScript support. See docs/extensions.md#tool-call-events. (#1147 by @giuseppeg)
  • Extension UI Protocol: Full RPC documentation and examples for extension dialogs and notifications, enabling headless clients to support interactive extensions. See docs/rpc.md#extension-ui-protocol. (#1144 by @aliou)

新增10

  • Added Linux ARM64 musl (Alpine Linux) support via clipboard dependency update
  • Added Android/Termux support with graceful clipboard fallback (#1164)
  • Added bash tool spawn hook support for adjusting command, cwd, and env before execution (#1160 by @mitsuhiko)
  • Added typed ToolCallEvent.input per tool with isToolCallEventType() type guard for narrowing built-in tool events (#1147 by @giuseppeg)
  • Exported discoverAndLoadExtensions from package to enable extension testing without a local repo clone (#1148)
  • Added Extension UI Protocol documentation to RPC docs covering all request/response types for extension dialogs and notifications (#1144 by @aliou)
  • Added rpc-demo.ts example extension exercising all RPC-supported extension UI methods (#1144 by @aliou)
  • Added rpc-extension-ui.ts TUI example client demonstrating the extension UI protocol with interactive dialogs (#1144 by @aliou)
  • Added PI_PACKAGE_DIR environment variable to override package path for content-addressed package managers (Nix, Guix) where store paths tokenize poorly (#1153 by @odysseus0)
  • /resume session picker now supports named-only filter toggle (default Ctrl+N, configurable via toggleSessionNamedFilter) to show only named sessions (#1128 by @w-winter)

修复8

  • Fixed pi update not updating npm/git packages when called without arguments (#1151)
  • Fixed models.json validation requiring fields documented as optional. Model definitions now only require id; all other fields (name, reasoning, input, cost, contextWindow, maxTokens) have sensible defaults. (#1146)
  • Fixed models resolving relative paths in skill files from cwd instead of skill directory by adding explicit guidance to skills preamble (#1136)
  • Fixed tree selector losing focus state when navigating entries (#1142 by @Perlence)
  • Fixed cacheRetention option not being passed through in buildBaseOptions (#1154)
  • Fixed OAuth login/refresh not using HTTP proxy settings (HTTP_PROXY, HTTPS_PROXY env vars) (#1132)
  • Fixed pi update <source> installing packages locally when the source is only registered globally (#1163 by @aliou)
  • Fixed tree navigation with summarization overwriting editor content typed during the summarization wait (#1169 by @aliou)
0.50.9 2026-02-01 新增 3 Added titlebar-spinner.ts example extension that shows a bra

新增3

  • Added titlebar-spinner.ts example extension that shows a braille spinner animation in the terminal title while the agent is working.
  • Added PI_AI_ANTIGRAVITY_VERSION environment variable documentation to help text (#1129)
  • Added cacheRetention stream option with provider-specific mappings for prompt cache controls, defaulting to short retention (#1134)
0.50.8 2026-02-01 新增 8修复 3 Added newSession, tree, and fork keybinding actions for /new

新增8

  • Added newSession, tree, and fork keybinding actions for /new, /tree, and /fork commands. All unbound by default. (#1114 by @juanibiapina)
  • Added retry.maxDelayMs setting to cap maximum server-requested retry delay. When a provider requests a longer delay (e.g., Google's "quota will reset after 5h"), the request fails immediately with an informative error instead of waiting silently. Default: 60000ms (60 seconds). (#1123)
  • /resume session picker: new "Threaded" sort mode (now default) displays sessions in a tree structure based on fork relationships. Compact one-line format with message count and age on the right. (#1124 by @pasky)
  • Added Qwen CLI OAuth provider extension example. (#940 by @4h9fbZ)
  • Added OAuth modifyModels hook support for extension-registered providers at registration time. (#940 by @4h9fbZ)
  • Added Qwen thinking format support for OpenAI-compatible completions via enable_thinking. (#940 by @4h9fbZ)
  • Added sticky column tracking for vertical cursor navigation so the editor restores the preferred column when moving across short lines. (#1120 by @Perlence)
  • Added resources_discover extension hook to supply additional skills, prompts, and themes on startup and reload.

修复3

  • Fixed switchSession() appending spurious thinking_level_change entry to session log on resume. setThinkingLevel() is now idempotent. (#1118)
  • Fixed clipboard image paste on WSL2/WSLg writing invalid PNG files when clipboard provides image/bmp format. BMP images are now converted to PNG before saving. (#1112 by @lightningRalf)
  • Fixed Kitty keyboard protocol base layout fallback so non-QWERTY layouts do not trigger wrong shortcuts (#1096 by @rytswd)
2026-0158 个版本
0.50.7 2026-01-31 修复 1 Multi-file extensions in packages now work correctly. Packag

修复1

  • Multi-file extensions in packages now work correctly. Package resolution now uses the same discovery logic as local extensions: only index.ts (or manifest-declared entries) are loaded from subdirectories, not helper modules. (#1102)
0.50.6 2026-01-30 新增 1修复 2 Added ctx.getSystemPrompt() to extension context for accessi

新增1

  • Added ctx.getSystemPrompt() to extension context for accessing the current effective system prompt (#1098 by @kaofelix)

修复2

  • Fixed empty rows appearing below footer when content shrinks (e.g., closing /tree, clearing multi-line editor) (#1095 by @marckrenn)
  • Fixed terminal cursor remaining hidden after exiting TUI via stop() when a render was pending (#1099 by @haoqixu)
0.50.5 2026-01-30 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.50.4 2026-01-30 新功能 8新增 8修复 5 OSC 52 clipboard support for SSH/mosh

新功能8

  • OSC 52 clipboard support for SSH/mosh - The /copy command now works over remote connections using the OSC 52 terminal escape sequence. No more clipboard frustration when using pi over SSH. (#1069 by @gturkoglu)
  • Vercel AI Gateway routing - Route requests through Vercel's AI Gateway with provider failover and load balancing. Configure via vercelGatewayRouting in models.json. (#1051 by @ben-vargas)
  • Character jump navigation - Bash/Readline-style character search: Ctrl+] jumps forward to the next occurrence of a character, Ctrl+Alt+] jumps backward. (#1074 by @Perlence)
  • Emacs-style Ctrl+B/Ctrl+F navigation - Alternative keybindings for word navigation (cursor word left/right) in the editor. (#1053 by @ninlds)
  • Line boundary navigation - Editor jumps to line start when pressing Up at first visual line, and line end when pressing Down at last visual line. (#1050 by @4h9fbZ)
  • Performance improvements - Optimized image line detection and box rendering cache in the TUI for better rendering performance. (#1084 by @can1357)
  • set_session_name RPC command - Headless clients can now set the session display name programmatically. (#1075 by @dnouri)
  • Disable double-escape behavior - New "none" option for doubleEscapeAction setting completely disables the double-escape shortcut. (#973 by @juanibiapina)

新增8

  • Added "none" option to doubleEscapeAction setting to disable double-escape behavior entirely (#973 by @juanibiapina)
  • Added OSC 52 clipboard support for SSH/mosh sessions. /copy now works over remote connections. (#1069 by @gturkoglu)
  • Added Vercel AI Gateway routing support via vercelGatewayRouting in models.json (#1051 by @ben-vargas)
  • Added Ctrl+B and Ctrl+F keybindings for cursor word left/right navigation in the editor (#1053 by @ninlds)
  • Added character jump navigation: Ctrl+] jumps forward to next character, Ctrl+Alt+] jumps backward (#1074 by @Perlence)
  • Editor now jumps to line start when pressing Up at first visual line, and line end when pressing Down at last visual line (#1050 by @4h9fbZ)
  • Optimized image line detection and box rendering cache for better TUI performance (#1084 by @can1357)
  • Added set_session_name RPC command for headless clients to set session display name (#1075 by @dnouri)

修复5

  • Read tool now handles macOS filenames with curly quotes (U+2019) and NFD Unicode normalization (#1078)
  • Respect .gitignore, .ignore, and .fdignore files when scanning package resources for skills, prompts, themes, and extensions (#1072)
  • Fixed tool call argument defaults when providers omit inputs (#1065)
  • Invalid JSON in settings.json no longer causes the file to be overwritten with empty settings (#1054)
  • Config selector now shows folder name for extensions with duplicate display names (#1064 by @Graffioh)
0.50.3 2026-01-29 新功能 1新增 1修复 1 Kimi For Coding provider

新功能1

  • Kimi For Coding provider: Access Moonshot AI's Anthropic-compatible coding API. Set KIMI_API_KEY environment variable. See README.md#kimi-for-coding.

新增1

  • Added Kimi For Coding provider support (Moonshot AI's Anthropic-compatible coding API). Set KIMI_API_KEY environment variable. See README.md#kimi-for-coding.

修复1

  • Resources now appear before messages when resuming a session, preventing loaded context from appearing at the bottom of the chat.
0.50.2 2026-01-29 新功能 5新增 6变更 2修复 15 Hugging Face provider

新功能5

  • Hugging Face provider: Access Hugging Face models via OpenAI-compatible Inference Router. Set HF_TOKEN environment variable. See README.md#hugging-face.
  • Extended prompt caching: PI_CACHE_RETENTION=long enables 1-hour caching for Anthropic (vs 5min default) and 24-hour for OpenAI (vs in-memory default). Only applies to direct API calls. See README.md#prompt-caching.
  • Configurable autocomplete height: autocompleteMaxVisible setting (3-20 items, default 5) controls dropdown size. Adjust via /settings or settings.json.
  • Shell-style keybindings: alt+b/alt+f for word navigation, ctrl+d for delete character forward. See docs/keybindings.md.
  • RPC get_commands: Headless clients can now list available commands programmatically. See docs/rpc.md.

新增6

  • Added Hugging Face provider support via OpenAI-compatible Inference Router (#994)
  • Added PI_CACHE_RETENTION environment variable to control cache TTL for Anthropic (5m vs 1h) and OpenAI (in-memory vs 24h). Set to long for extended retention. (#967)
  • Added autocompleteMaxVisible setting for configurable autocomplete dropdown height (3-20 items, default 5) (#972 by @masonc15)
  • Added /files command to list all file operations (read, write, edit) in the current session
  • Added shell-style keybindings: alt+b/alt+f for word navigation, ctrl+d for delete character forward (when editor has text) (#1043 by @jasonish)
  • Added get_commands RPC method for headless clients to list available commands (#995 by @dnouri)

变更2

  • Improved extractCursorPosition performance in TUI: scans lines in reverse order, early-outs when cursor is above viewport (#1004 by @can1357)
  • Autocomplete improvements: better handling of partial matches and edge cases (#1024 by @Perlence)

修复15

  • External edits to settings.json are now preserved when pi reloads or saves unrelated settings. Previously, editing settings.json directly (e.g., removing a package from packages array) would be silently reverted on next pi startup when automatic setters like setLastChangelogVersion() triggered a save.
  • Fixed custom header not displaying correctly with quietStartup enabled (#1039 by @tudoroancea)
  • Empty array in package filter now disables all resources instead of falling back to manifest defaults (#1044)
  • Auto-retry counter now resets after each successful LLM response instead of accumulating across tool-use turns (#1019)
  • Fixed incorrect .md file names in warning messages (#1041 by @llimllib)
  • Fixed provider name hidden in footer when terminal is narrow (#981 by @Perlence)
  • Fixed backslash input buffering causing delayed character display in editor (#1037 by @Perlence)
  • Fixed markdown table rendering with proper row dividers and minimum column width (#997 by @tmustier)
  • Fixed OpenAI completions toolChoice handling (#998 by @williamtwomey)
  • Fixed cross-provider handoff failing when switching from OpenAI Responses API providers due to pipe-separated tool call IDs (#1022)
  • Fixed 429 rate limit errors incorrectly triggering auto-compaction instead of retry with backoff (#1038)
  • Fixed Anthropic provider to handle sensitive stop_reason returned by API (#978)
  • Fixed DeepSeek API compatibility by detecting deepseek.com URLs and disabling unsupported developer role (#1048)
  • Fixed Anthropic provider to preserve input token counts when proxies omit them in message_delta events (#1045)
  • Fixed autocompleteMaxVisible setting not persisting to settings.json
0.50.1 2026-01-26 修复 6 Git extension updates now handle force-pushed remotes gracef

修复6

  • Git extension updates now handle force-pushed remotes gracefully instead of failing (#961 by @aliou)
  • Extension ctx.newSession({ setup }) now properly syncs agent state and renders messages after setup callback runs (#968)
  • Fixed extension UI bindings not initializing when starting with no extensions, which broke UI methods after /reload
  • Fixed /hotkeys output to title-case extension hotkeys (#969 by @Perlence)
  • Fixed model catalog generation to exclude deprecated OpenCode Zen models (#970 by @DanielTatarkin)
  • Fixed git extension removal to prune empty directories
0.50.0 2026-01-26 新功能 9破坏性变更 4新增 23变更 6修复 33 Pi packages for bundling and installing extensions, skills,

新功能9

  • Pi packages for bundling and installing extensions, skills, prompts, and themes. See docs/packages.md.
  • Hot reload (/reload) of resources including AGENTS.md, SYSTEM.md, APPEND_SYSTEM.md, prompt templates, skills, themes, and extensions. See README.md#commands and README.md#context-files.
  • Custom providers via pi.registerProvider() for proxies, custom endpoints, OAuth or SSO flows, and non-standard streaming APIs. See docs/custom-provider.md.
  • Azure OpenAI Responses provider support with deployment-aware model mapping. See docs/providers.md#azure-openai.
  • OpenRouter routing support for custom models via openRouterRouting. See docs/providers.md#api-keys and docs/models.md.
  • Skill invocation messages are now collapsible and skills can opt out of model invocation via disable-model-invocation. See docs/skills.md#frontmatter.
  • Session selector renaming and configurable keybindings. See README.md#commands and docs/keybindings.md.
  • models.json headers can resolve environment variables and shell commands. See docs/models.md#value-resolution.
  • --verbose CLI flag to override quiet startup. See README.md#cli-reference.

破坏性变更4

  • Header values in models.json now resolve environment variables (if a header value matches an env var name, the env var value is used). This may change behavior if a literal header value accidentally matches an env var name. (#909)
  • External packages (npm/git) are now configured via packages array in settings.json instead of extensions. Existing npm:/git: entries in extensions are auto-migrated. (#645)
  • Resource loading now uses ResourceLoader only and settings.json uses arrays for extensions, skills, prompts, and themes (#645)
  • Removed discoverAuthStorage and discoverModels from the SDK. AuthStorage and ModelRegistry now default to ~/.pi/agent paths unless you pass an agentDir (#645)

新增23

  • Session renaming in /resume picker via Ctrl+R without opening the session (#863 by @svkozak)
  • Session selector keybindings are now configurable (#948 by @aos)
  • disable-model-invocation frontmatter field for skills to prevent agentic invocation while still allowing explicit /skill:name commands (#927)
  • Exposed copyToClipboard utility for extensions (#926 by @mitsuhiko)
  • Skill invocation messages are now collapsible in chat output, showing collapsed by default with skill name and expand hint (#894)
  • Header values in models.json now support environment variables and shell commands, matching apiKey resolution (#909)
  • Added HTTP proxy environment variable support for API requests (#942 by @haoqixu)
  • Added OpenRouter provider routing support for custom models via openRouterRouting compat field (#859 by @v01dpr1mr0s3)
  • Added azure-openai-responses provider support for Azure OpenAI Responses API. (#890 by @markusylisiurunen)
  • Added changelog link to update notifications (#925 by @dannote)
  • Added --verbose CLI flag to override quietStartup setting (#906 by @Perlence)
  • markdown.codeBlockIndent setting to customize code block indentation in rendered output
  • Extension package management with pi install, pi remove, pi update, and pi list commands (#645)
  • Package filtering: selectively load resources from packages using object form in packages array (#645)
  • Glob pattern support with minimatch in package filters, top-level settings arrays, and pi manifest (e.g., "!funky.json", "*.ts") (#645)
  • /reload command to reload extensions, skills, prompts, and themes (#645)
  • pi config command with TUI to enable/disable package and top-level resources via patterns (#938)
  • CLI flags for --skill, --prompt-template, --theme, --no-prompt-templates, and --no-themes (#645)
  • Package deduplication: if same package appears in global and project settings, project wins (#645)
  • Unified collision reporting with ResourceDiagnostic type for all resource types (#645)
  • Show provider alongside the model in the footer if multiple providers are available
  • Custom provider support via pi.registerProvider() with streamSimple for custom API implementations
  • Added custom-provider.ts example extension demonstrating custom Anthropic provider with OAuth

变更6

  • /resume picker sort toggle moved to Ctrl+S to free Ctrl+R for rename (#863 by @svkozak)
  • HTML export: clicking a sidebar message now navigates to its newest leaf and scrolls to it, instead of truncating the branch (#853 by @mitsuhiko)
  • HTML export: active path is now visually highlighted with dimmed off-path nodes (#929 by @hewliyang)
  • Azure OpenAI Responses provider now uses base URL configuration with deployment-aware model mapping and no longer includes service tier handling
  • /reload now re-renders the entire scrollback so updated extension components are visible immediately (#928 by @ferologics)
  • Skill, prompt template, and theme discovery now use settings and CLI path arrays instead of legacy filters (#645)

修复33

  • Extension setWorkingMessage() calls in agent_start handlers now work correctly; previously the message was silently ignored because the loading animation didn't exist yet (#935)
  • Fixed package auto-discovery to respect loader rules, config overrides, and force-exclude patterns
  • Fixed /reload restoring the correct editor after reload (#949 by @Perlence)
  • Fixed distributed themes breaking /export (#946 by @mitsuhiko)
  • Fixed startup hints to clarify thinking level selection and expanded thinking guidance
  • Fixed SDK initial model resolution to use findInitialModel and default to Claude Opus 4.5 for Anthropic models
  • Fixed no-models warning to include the /model instruction
  • Fixed authentication error messages to point to the authentication documentation
  • Fixed bash output hint lines to truncate to terminal width
  • Fixed custom editors to honor the paddingX setting (#936 by @Perlence)
  • Fixed system prompt tool list to show only built-in tools
  • Fixed package manager to check npm package versions before using cached copies
  • Fixed package manager to run npm install after cloning git repositories with a package.json
  • Fixed extension provider registrations to apply before model resolution
  • Fixed editor multi-line insertion handling and lastAction tracking (#945 by @Perlence)
  • Fixed editor word wrapping to reserve a cursor column (#934 by @Perlence)
  • Fixed editor word wrapping to use single-pass backtracking for whitespace handling (#924 by @Perlence)
  • Fixed Kitty image ID allocation and cleanup to prevent image ID collisions
  • Fixed overlays staying centered after terminal resizes (#950 by @nicobailon)
  • Fixed streaming dispatch to use the model api type instead of hardcoded API defaults
  • Fixed Google providers to default tool call arguments to an empty object when omitted
  • Fixed OpenAI Responses streaming to handle arguments.done events on OpenAI-compatible endpoints (#917 by @williballenthin)
  • Fixed OpenAI Codex Responses tool strictness handling after the shared responses refactor
  • Fixed Azure OpenAI Responses streaming to guard deltas before content parts and correct metadata and handoff gating
  • Fixed OpenAI completions tool-result image batching after consecutive tool results (#902 by @terrorobe)
  • Off-by-one error in bash output "earlier lines" count caused by counting spacing newline as hidden content (#921)
  • User package filters now layer on top of manifest filters instead of replacing them (#645)
  • Auto-retry now handles "terminated" errors from Codex API mid-stream failures
  • Follow-up queue (Alt+Enter) now sends full paste content instead of [paste #N ...] markers (#912)
  • Fixed Alt-Up not restoring messages queued during compaction (#923 by @aliou)
  • Fixed session corruption when loading empty or invalid session files via --session flag (#932 by @armanddp)
  • Fixed extension shortcuts not firing when extension also uses setEditorComponent() (#947 by @Perlence)
  • Session "modified" time now uses last message timestamp instead of file mtime, so renaming doesn't reorder the recent list (#863 by @svkozak)
0.49.3 2026-01-22 新增 5变更 2修复 9 markdown.codeBlockIndent setting to customize code block ind

新增5

  • markdown.codeBlockIndent setting to customize code block indentation in rendered output (#855 by @terrorobe)
  • Added inline-bash.ts example extension for expanding !{command} patterns in prompts (#881 by @scutifer)
  • Added antigravity-image-gen.ts example extension for AI image generation via Google Antigravity (#893 by @ben-vargas)
  • Added PI_SHARE_VIEWER_URL environment variable for custom share viewer URLs (#889 by @andresaraujo)
  • Added Alt+Delete as hotkey for delete word forwards (#878 by @Perlence)

变更2

  • Tree selector: changed label filter shortcut from l to Shift+L so users can search for entries containing "l" (#861 by @mitsuhiko)
  • Fuzzy matching now scores consecutive matches higher for better search relevance (#860 by @mitsuhiko)

修复9

  • Fixed error messages showing hardcoded ~/.pi/agent/ paths instead of respecting PI_CODING_AGENT_DIR (#887 by @aliou)
  • Fixed write tool not displaying errors in the UI when execution fails (#856)
  • Fixed HTML export using default theme instead of user's active theme (#870 by @scutifer)
  • Show session name in the footer and terminal / tab title (#876 by @scutifer)
  • Fixed 256color fallback in Terminal.app to prevent color rendering issues (#869 by @Perlence)
  • Fixed viewport tracking and cursor positioning for overlays and content shrink scenarios
  • Fixed autocomplete to allow searches with / characters (e.g., folder1/folder2) (#882 by @richardgill)
  • Fixed autolinked emails displaying redundant (mailto:...) suffix (#888 by @terrorobe)
  • Fixed @ file autocomplete adding space after directories, breaking continued autocomplete into subdirectories
0.49.2 2026-01-19 新增 3变更 2修复 3移除 1 Added widget placement option for extension widgets via widg

新增3

  • Added widget placement option for extension widgets via widgetPlacement in pi.addWidget() (#850 by @marckrenn)
  • Added AWS credential detection for ECS/Kubernetes environments: AWS_CONTAINER_CREDENTIALS_RELATIVE_URI, AWS_CONTAINER_CREDENTIALS_FULL_URI, AWS_WEB_IDENTITY_TOKEN_FILE (#848)
  • Add "quiet startup" setting to /settings (#847 by @unexge)

变更2

  • HTML export now includes JSONL download button, jump-to-last-message on click, and fixed missing labels (#853 by @mitsuhiko)
  • Improved error message for OAuth authentication failures (expired credentials, offline) instead of generic 'No API key found' (#849 by @zedrdave)

修复3

  • Fixed /model selector scope toggle so you can switch between all and scoped models when scoped models are saved (#844)
  • Fixed OpenAI Responses 400 error "reasoning without following item" when replaying aborted turns (#838)
  • Fixed pi exiting with code 0 when cancelling resume session selection

移除1

  • Removed strictResponsesPairing compat option from models.json schema (no longer needed)
0.49.1 2026-01-18 新增 3变更 2修复 4 Added strictResponsesPairing compat option for custom OpenAI

新增3

  • Added strictResponsesPairing compat option for custom OpenAI Responses models on Azure (#768 by @prateekmedia)
  • Session selector (/resume) now supports path display toggle (Ctrl+P) and session deletion (Ctrl+D) with inline confirmation (#816 by @w-winter)
  • Added undo support in interactive mode with Ctrl+- hotkey. (#831 by @Perlence)

变更2

  • Share URLs now use hash fragments (#) instead of query strings (?) to prevent session IDs from being sent to buildwithpi.ai (#829 by @terrorobe)
  • API keys in models.json can now be retrieved via shell command using ! prefix (e.g., "apiKey": "!security find-generic-password -ws 'anthropic'" for macOS Keychain) (#762 by @cv)

修复4

  • Fixed IME candidate window appearing in wrong position when filtering menus with Input Method Editor (e.g., Chinese IME). Components with search inputs now properly propagate focus state for cursor positioning. (#827)
  • Fixed extension shortcut conflicts to respect user keybindings when built-in actions are remapped. (#826 by @richardgill)
  • Fixed photon WASM loading in standalone compiled binaries.
  • Fixed tool call ID normalization for cross-provider handoffs (e.g., Codex to Antigravity Claude) (#821)
0.49.0 2026-01-17 新增 7变更 2修复 5移除 1 pi.setLabel(entryId, label) in ExtensionAPI for setting per-

新增7

  • pi.setLabel(entryId, label) in ExtensionAPI for setting per-entry labels from extensions (#806)
  • Export keyHint, appKeyHint, editorKey, appKey, rawKeyHint for extensions to format keybinding hints consistently (#802 by @dannote)
  • Exported VERSION from the package index and updated the custom-header example. (#798 by @tallshort)
  • Added showHardwareCursor setting to control cursor visibility while still positioning it for IME support. (#800 by @ghoulr)
  • Added Emacs-style kill ring editing with yank and yank-pop keybindings, plus legacy Alt+letter handling and Alt+D delete word forward support in the interactive editor. (#810 by @Perlence)
  • Added ctx.compact() and ctx.getContextUsage() to extension contexts for programmatic compaction and context usage checks.
  • Added documentation for delete word forward and kill ring keybindings in interactive mode. (#810 by @Perlence)

变更2

  • Updated the default system prompt wording to clarify the pi harness and documentation scope.
  • Simplified Codex system prompt handling to use the default system prompt directly for Codex instructions.

修复5

  • Fixed photon module failing to load in ESM context with "require is not defined" error (#795 by @dannote)
  • Fixed compaction UI not showing when extensions trigger compaction.
  • Fixed orphaned tool results after errored assistant messages causing Codex API errors. When an assistant message has stopReason: "error", its tool calls are now excluded from pending tool tracking, preventing synthetic tool results from being generated for calls that will be dropped by provider-specific converters. (#812)
  • Fixed Bedrock Claude max_tokens handling to always exceed thinking budget tokens, preventing compaction failures. (#797 by @pjtf93)
  • Fixed Claude Code tool name normalization to match the Claude Code tool list case-insensitively and remove invalid mappings.

移除1

  • Removed pi-internal:// path resolution from the read tool.
0.48.0 2026-01-16 新增 9修复 10变更 1 Added quietStartup setting to silence startup output (versio

新增9

  • Added quietStartup setting to silence startup output (version header, loaded context info, model scope line). Changelog notifications are still shown. (#777 by @ribelo)
  • Added editorPaddingX setting for horizontal padding in input editor (0-3, default: 0)
  • Added shellCommandPrefix setting to prepend commands to every bash execution, enabling alias expansion in non-interactive shells (e.g., "shellCommandPrefix": "shopt -s expand_aliases") (#790 by @richardgill)
  • Added bash-style argument slicing for prompt templates (#770 by @airtonix)
  • Extension commands can provide argument auto-completions via getArgumentCompletions in pi.registerCommand() (#775 by @ribelo)
  • Bash tool now displays the timeout value in the UI when a timeout is set (#780 by @dannote)
  • Export getShellConfig for extensions to detect user's shell environment (#766 by @dannote)
  • Added thinkingText and selectedBg to theme schema (#763 by @scutifer)
  • navigateTree() now supports replaceInstructions option to replace the default summarization prompt entirely, and label option to attach a label to the branch summary entry (#787 by @mitsuhiko)

修复10

  • Fixed crash during auto-compaction when summarization fails (e.g., quota exceeded). Now displays error message instead of crashing (#792)
  • Fixed --session <UUID> to search globally across projects if not found locally, with option to fork sessions from other projects (#785 by @ribelo)
  • Fixed standalone binary WASM loading on Linux (#784)
  • Fixed string numbers in tool arguments not being coerced to numbers during validation (#786 by @dannote)
  • Fixed --no-extensions flag not preventing extension discovery (#776)
  • Fixed extension messages rendering twice on startup when pi.sendMessage({ display: true }) is called during session_start (#765 by @dannote)
  • Fixed PI_CODING_AGENT_DIR env var not expanding tilde (~) to home directory (#778 by @aliou)
  • Fixed session picker hint text overflow (#764)
  • Fixed Kitty keyboard protocol shifted symbol keys (e.g., @, ?) not working in editor (#779 by @iamd3vil)
  • Fixed Bedrock tool call IDs causing API errors from invalid characters (#781 by @pjtf93)

变更1

  • Hardware cursor is now disabled by default for better terminal compatibility. Set PI_HARDWARE_CURSOR=1 to enable (replaces PI_NO_HARDWARE_CURSOR=1 which disabled it).
0.47.0 2026-01-16 破坏性变更 1新增 6变更 1修复 7 Extensions using Editor directly must now pass TUI as the fi

破坏性变更1

  • Extensions using Editor directly must now pass TUI as the first constructor argument: new Editor(tui, theme). The tui parameter is available in extension factory functions. (#732)

新增6

  • OpenAI Codex official support: Full compatibility with OpenAI's Codex CLI models (gpt-5.1, gpt-5.2, gpt-5.1-codex-mini, gpt-5.2-codex). Features include static system prompt for OpenAI allowlisting, prompt caching via session ID, and reasoning signature retention across turns. Set OPENAI_API_KEY and use --provider openai-codex or select a Codex model. (#737)
  • pi-internal:// URL scheme in read tool for accessing internal documentation. The model can read files from the coding-agent package (README, docs, examples) to learn about extending pi.
  • New input event in extension system for intercepting, transforming, or handling user input before the agent processes it. Supports three result types: continue (pass through), transform (modify text/images), handled (respond without LLM). Handlers chain transforms and short-circuit on handled. (#761 by @nicobailon)
  • Extension example: input-transform.ts demonstrating input interception patterns (quick mode, instant commands, source routing) (#761 by @nicobailon)
  • Custom tool HTML export: extensions with renderCall/renderResult now render in /share and /export output with ANSI-to-HTML color conversion (#702 by @aliou)
  • Direct filter shortcuts in Tree mode: Ctrl+D (default), Ctrl+T (no-tools), Ctrl+U (user-only), Ctrl+L (labeled-only), Ctrl+A (all) (#747 by @kaofelix)

变更1

  • Skill commands (/skill:name) are now expanded in AgentSession instead of interactive mode. This enables skill commands in RPC and print modes, and allows the input event to intercept /skill:name before expansion.

修复7

  • Editor no longer corrupts terminal display when loading large prompts via setEditorText. Content now scrolls vertically with indicators showing lines above/below the viewport. (#732)
  • Piped stdin now works correctly: echo foo | pi is equivalent to pi -p foo. When stdin is piped, print mode is automatically enabled since interactive mode requires a TTY (#708)
  • Session tree now preserves branch connectors and indentation when filters hide intermediate entries so descendants attach to the nearest visible ancestor and sibling branches align. Fixed in both TUI and HTML export (#739 by @w-winter)
  • Added upstream connect, connection refused, and reset before headers patterns to auto-retry error detection (#733)
  • Multi-line YAML frontmatter in skills and prompt templates now parses correctly. Centralized frontmatter parsing using the yaml library. (#728 by @richardgill)
  • ctx.shutdown() now waits for pending UI renders to complete before exiting, ensuring notifications and final output are visible (#756)
  • OpenAI Codex provider now retries on transient errors (429, 5xx, connection failures) with exponential backoff (#733)
0.46.0 2026-01-15 修复 8新增 7变更 4 Scoped models (--models or enabledModels) now remember the l

修复8

  • Scoped models (--models or enabledModels) now remember the last selected model across sessions instead of always starting with the first model in the scope (#736 by @ogulcancelik)
  • Show bun install instead of npm install in update notification when running under Bun (#714 by @dannote)
  • /skill prompts now include the skill path (#711 by @jblwilliams)
  • Use configurable expandTools keybinding instead of hardcoded Ctrl+O (#717 by @dannote)
  • Compaction turn prefix summaries now merge correctly (#738 by @vsabavat)
  • Avoid unsigned Gemini 3 tool calls (#741 by @roshanasingh4)
  • Fixed signature support for non-Anthropic models in Amazon Bedrock provider (#727 by @unexge)
  • Keyboard shortcuts (Ctrl+C, Ctrl+D, etc.) now work on non-Latin keyboard layouts (Russian, Ukrainian, Bulgarian, etc.) in terminals supporting Kitty keyboard protocol with alternate key reporting (#718 by @dannote)

新增7

  • Edit tool now uses fuzzy matching as fallback when exact match fails, tolerating trailing whitespace, smart quotes, Unicode dashes, and special spaces (#713 by @dannote)
  • Support APPEND_SYSTEM.md to append instructions to the system prompt (#716 by @tallshort)
  • Session picker search: Ctrl+R toggles sorting between fuzzy match (default) and most recent; supports quoted phrase matching and re: regex mode (#731 by @ogulcancelik)
  • Export getAgentDir for extensions (#749 by @dannote)
  • Show loaded prompt templates on startup (#743 by @tallshort)
  • MiniMax China (minimax-cn) provider support (#725 by @tallshort)
  • gpt-5.2-codex models for GitHub Copilot and OpenCode Zen providers (#734 by @aadishv)

变更4

  • Replaced wasm-vips with @silvia-odwyer/photon-node for image processing (#710 by @can1357)
  • Extension example: plan-mode/ shortcut changed from Shift+P to Ctrl+Alt+P to avoid conflict with typing capital P (#746 by @ferologics)
  • UI keybinding hints now respect configured keybindings across components (#724 by @dannote)
  • CLI process title is now set to pi for easier process identification (#742 by @richardgill)
0.45.7 2026-01-13 新增 1 Exported highlightCode and getLanguageFromPath for extension

新增1

  • Exported highlightCode and getLanguageFromPath for extensions (#703 by @dannote)
0.45.6 2026-01-13 新增 4 ctx.ui.custom() now accepts overlayOptions for overlay posit

新增4

  • ctx.ui.custom() now accepts overlayOptions for overlay positioning and sizing (anchor, margins, offsets, percentages, absolute positioning) (#667 by @nicobailon)
  • ctx.ui.custom() now accepts onHandle callback to receive the OverlayHandle for controlling overlay visibility (#667 by @nicobailon)
  • Extension example: overlay-qa-tests.ts with 10 commands for testing overlay positioning, animation, and toggle scenarios (#667 by @nicobailon)
  • Extension example: doom-overlay/ - DOOM game running as an overlay at 35 FPS (auto-downloads WAD on first run) (#667 by @nicobailon)
0.45.5 2026-01-13 修复 1 Skip changelog display on fresh install (only show on upgrad

修复1

  • Skip changelog display on fresh install (only show on upgrades)
0.45.4 2026-01-13 变更 2新增 5修复 4 Light theme colors adjusted for WCAG AA compliance (4.5:1 co

变更2

  • Light theme colors adjusted for WCAG AA compliance (4.5:1 contrast ratio against white backgrounds)
  • Replaced sharp with wasm-vips for image processing (resize, PNG conversion). Eliminates native build requirements that caused installation failures on some systems. (#696)

新增5

  • Extension example: summarize.ts for summarizing conversations using custom UI and an external model (#684 by @scutifer)
  • Extension example: question.ts enhanced with custom UI for asking user questions (#693 by @ferologics)
  • Extension example: plan-mode/ enhanced with explicit step tracking and progress widget (#694 by @ferologics)
  • Extension example: questionnaire.ts for multi-question input with tab bar navigation (#695 by @ferologics)
  • Experimental Vercel AI Gateway provider support: set AI_GATEWAY_API_KEY and use --provider vercel-ai-gateway. Token usage is currently reported incorrectly by Anthropic Messages compatible endpoint. (#689 by @timolins)

修复4

  • Fix API key resolution after model switches by using provider argument (#691 by @joshp123)
  • Fixed z.ai thinking/reasoning: thinking toggle now correctly enables/disables thinking for z.ai models (#688)
  • Fixed extension loading in compiled Bun binary: extensions with local file imports now work correctly. Updated @mariozechner/jiti to v2.6.5 which bundles babel for Bun binary compatibility. (#681)
  • Fixed theme loading when installed via mise: use wrapper directory in release tarballs for compatibility with mise's strip_components=1 extraction. (#681)
0.45.3 2026-01-13 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.45.2 2026-01-13 修复 1 Extensions now load correctly in compiled Bun binary using @

修复1

  • Extensions now load correctly in compiled Bun binary using @mariozechner/jiti fork with virtualModules support. Bundled packages (@sinclair/typebox, @mariozechner/pi-tui, @mariozechner/pi-ai, @mariozechner/pi-coding-agent) are accessible to extensions without filesystem node_modules.
0.45.1 2026-01-13 变更 1 /share now outputs buildwithpi.ai session preview URLs inste

变更1

  • /share now outputs buildwithpi.ai session preview URLs instead of pi.dev
0.45.0 2026-01-13 新增 5 MiniMax provider support: set MINIMAX_API_KEY and use minima

新增5

  • MiniMax provider support: set MINIMAX_API_KEY and use minimax/MiniMax-M2.1 (#656 by @dannote)
  • /scoped-models: Alt+Up/Down to reorder enabled models. Order is preserved when saving with Ctrl+S and determines Ctrl+P cycling order. (#676 by @thomasmhr)
  • Amazon Bedrock provider support (experimental, tested with Anthropic Claude models only) (#494 by @unexge)
  • Extension example: sandbox/ for OS-level bash sandboxing using @anthropic-ai/sandbox-runtime with per-project config (#673 by @dannote)
  • Print mode JSON output now emits the session header as the first line.
0.44.0 2026-01-12 破坏性变更 1新增 5修复 2 pi.getAllTools() now returns ToolInfo[] (with name and descr

破坏性变更1

  • pi.getAllTools() now returns ToolInfo[] (with name and description) instead of string[]. Extensions that only need names can use .map(t => t.name). (#648 by @carsonfarmer)

新增5

  • Session naming: /name <name> command sets a display name shown in the session selector instead of the first message. Useful for distinguishing forked sessions. Extensions can use pi.setSessionName() and pi.getSessionName(). (#650 by @scutifer)
  • Extension example: notify.ts for desktop notifications via OSC 777 escape sequence (#658 by @ferologics)
  • Inline hint for queued messages showing the Alt+Up restore shortcut (#657 by @tmustier)
  • Page-up/down navigation in /resume session selector to jump by 5 items (#662 by @aliou)
  • Fuzzy search in /settings menu: type to filter settings by label (#643 by @ninlds)

修复2

  • Session selector now stays open when current folder has no sessions, allowing Tab to switch to "all" scope (#661 by @aliou)
  • Extensions using theme utilities like getSettingsListTheme() now work in dev mode with tsx
0.43.0 2026-01-11 破坏性变更 3新增 10修复 5 Extension editor (ctx.ui.editor()) now uses Enter to submit

破坏性变更3

  • Extension editor (ctx.ui.editor()) now uses Enter to submit and Shift+Enter for newlines, matching the main editor. Previously used Ctrl+Enter to submit. Extensions with hardcoded "ctrl+enter" hints need updating. (#642 by @mitsuhiko)
  • Renamed /branch command to /fork (#641)

    · RPC: branch → fork, get_branch_messages → get_fork_messages

    · SDK: branch() → fork(), getBranchMessages() → getForkMessages()

    · AgentSession: branch() → fork(), getUserMessagesForBranching() → getUserMessagesForForking()

    · Extension events: session_before_branch → session_before_fork, session_branch → session_fork

    · Settings: doubleEscapeAction: "branch" | "tree" → "fork" | "tree"

  • SessionManager.list() and SessionManager.listAll() are now async, returning Promise<SessionInfo[]>. Callers must await them. (#620 by @tmustier)

新增10

  • /resume selector now toggles between current-folder and all sessions with Tab, showing the session cwd in the All view and loading progress. (#620 by @tmustier)
  • SessionManager.list() and SessionManager.listAll() accept optional onProgress callback for progress updates
  • SessionInfo.cwd field containing the session's working directory (empty string for old sessions)
  • SessionListProgress type export for progress callbacks
  • /scoped-models command to enable/disable models for Ctrl+P cycling. Changes are session-only by default; press Ctrl+S to persist to settings.json. (#626 by @CarlosGtrz)
  • model_select extension hook fires when model changes via /model, model cycling, or session restore with source field and previousModel (#628 by @marckrenn)
  • ctx.ui.setWorkingMessage() extension API to customize the "Working..." message during streaming (#625 by @nicobailon)
  • Skill slash commands: loaded skills are registered as /skill:name commands for quick access. Toggle via /settings or skills.enableSkillCommands in settings.json. (#630 by @Dwsy)
  • Slash command autocomplete now uses fuzzy matching (type /skbra to match /skill:brave-search)
  • /tree branch summarization now offers three options: "No summary", "Summarize", and "Summarize with custom prompt". Custom prompts are appended as additional focus to the default summarization instructions. (#642 by @mitsuhiko)

修复5

  • Missing spacer between assistant message and text editor (#655)
  • Session picker respects custom keybindings when using --resume (#633 by @aos)
  • Custom footer extensions now see model changes: ctx.model is now a getter that returns the current model instead of a snapshot from when the context was created (#634 by @ogulcancelik)
  • Footer git branch not updating after external branch switches. Git uses atomic writes (temp file + rename), which changes the inode and breaks fs.watch on the file. Now watches the directory instead.
  • Extension loading errors are now displayed to the user instead of being silently ignored (#639 by @aliou)
0.42.5 2026-01-11 修复 4 Reduced flicker by only re-rendering changed lines (#617 by

修复4

  • Reduced flicker by only re-rendering changed lines (#617 by @ogulcancelik). No worries tho, there's still a little flicker in the VS Code Terminal. Praise the flicker.
  • Cursor position tracking when content shrinks with unchanged remaining lines
  • TUI renders with wrong dimensions after suspend/resume if terminal was resized while suspended (#599)
  • Pasted content containing Kitty key release patterns (e.g., :3F in MAC addresses) was incorrectly filtered out (#623 by @ogulcancelik)
0.42.4 2026-01-10 修复 2 Bash output expanded hint now says "(ctrl+o to collapse)" (#

修复2

  • Bash output expanded hint now says "(ctrl+o to collapse)" (#610 by @tallshort)
  • Fixed UTF-8 text corruption in remote bash execution (SSH, containers) by using streaming TextDecoder (#608)
0.42.3 2026-01-10 变更 1 OpenAI Codex: updated to use bundled system prompt from upst

变更1

  • OpenAI Codex: updated to use bundled system prompt from upstream
0.42.2 2026-01-10 新增 3修复 1 /model <search> now pre-filters the model selector or auto-s

新增3

  • /model <search> now pre-filters the model selector or auto-selects on exact match. Use provider/model syntax to disambiguate (e.g., /model openai/gpt-4). (#587 by @zedrdave)
  • FooterDataProvider for custom footers: ctx.ui.setFooter() now receives a third footerData parameter providing getGitBranch(), getExtensionStatuses(), and onBranchChange() for reactive updates (#600 by @nicobailon)
  • Alt+Up hotkey to restore queued steering/follow-up messages back into the editor without aborting the current run (#604 by @tmustier)

修复1

  • Fixed LM Studio compatibility for OpenAI Responses tool strict mapping in the ai provider (#598 by @gnattu)
0.42.1 2026-01-09 修复 1 Symlinked directories in prompts/ folders are now followed w

修复1

  • Symlinked directories in prompts/ folders are now followed when loading prompt templates (#601 by @aliou)
0.42.0 2026-01-09 新增 1 Added OpenCode Zen provider support. Set OPENCODE_API_KEY en

新增1

  • Added OpenCode Zen provider support. Set OPENCODE_API_KEY env var and use opencode/<model-id> (e.g., opencode/claude-opus-4-5).
0.41.0 2026-01-09 新增 1 Anthropic OAuth support is back! Use /login to authenticate

新增1

  • Anthropic OAuth support is back! Use /login to authenticate with your Claude Pro/Max subscription.
0.40.1 2026-01-09 移除 1 Anthropic OAuth support (/login). Use API keys instead.

移除1

  • Anthropic OAuth support (/login). Use API keys instead.
0.40.0 2026-01-08 新增 1修复 1 Documentation on component invalidation and theme changes in

新增1

  • Documentation on component invalidation and theme changes in docs/tui.md

修复1

  • Components now properly rebuild their content on theme change (tool executions, assistant messages, bash executions, custom messages, branch/compaction summaries)
0.39.1 2026-01-08 修复 2 setTheme() now triggers a full rerender so previously render

修复2

  • setTheme() now triggers a full rerender so previously rendered components update with the new theme colors
  • mac-system-theme.ts example now polls every 2 seconds and uses osascript for real-time macOS appearance detection
0.39.0 2026-01-08 破坏性变更 2新增 11修复 7 before_agent_start event now receives systemPrompt in the ev

破坏性变更2

  • before_agent_start event now receives systemPrompt in the event object and returns systemPrompt (full replacement) instead of systemPromptAppend. Extensions that were appending must now use event.systemPrompt + extra pattern. (#575)
  • discoverSkills() now returns { skills: Skill[], warnings: SkillWarning[] } instead of Skill[]. This allows callers to handle skill loading warnings. (#577 by @cv)

新增11

  • ctx.ui.getAllThemes(), ctx.ui.getTheme(name), and ctx.ui.setTheme(name | Theme) methods for extensions to list, load, and switch themes at runtime (#576)
  • --no-tools flag to disable all built-in tools, allowing extension-only tool setups (#557 by @cv)
  • Pluggable operations for built-in tools enabling remote execution via SSH or other transports (#564). Interfaces: ReadOperations, WriteOperations, EditOperations, BashOperations, LsOperations, GrepOperations, FindOperations
  • user_bash event for intercepting user !/!! commands, allowing extensions to redirect to remote systems (#528)
  • setActiveTools() in ExtensionAPI for dynamic tool management
  • Built-in renderers used automatically for tool overrides without custom renderCall/renderResult
  • ssh.ts example: remote tool execution via --ssh user@host:/path
  • interactive-shell.ts example: run interactive commands (vim, git rebase, htop) with full terminal access via !i prefix or auto-detection
  • Wayland clipboard support for /copy command using wl-copy with xclip/xsel fallback (#570 by @OgulcanCelik)
  • Experimental: ctx.ui.custom() now accepts { overlay: true } option for floating modal components that composite over existing content without clearing the screen (#558 by @nicobailon)
  • AgentSession.skills and AgentSession.skillWarnings properties to access loaded skills without rediscovery (#577 by @cv)

修复7

  • String systemPrompt in createAgentSession() now works as a full replacement instead of having context files and skills appended, matching documented behavior (#543)
  • Update notification for bun binary installs now shows release download URL instead of npm command (#567 by @ferologics)
  • ESC key now works during "Working..." state after auto-retry (#568 by @tmustier)
  • Abort messages now show correct retry attempt count (e.g., "Aborted after 2 retry attempts") (#568 by @tmustier)
  • Fixed Antigravity provider returning 429 errors despite available quota (#571 by @ben-vargas)
  • Fixed malformed thinking text in Gemini/Antigravity responses where thinking content appeared as regular text or vice versa. Cross-model conversations now properly convert thinking blocks to plain text. (#561)
  • --no-skills flag now correctly prevents skills from loading in interactive mode (#577 by @cv)
0.38.0 2026-01-08 破坏性变更 7新增 8修复 7 ctx.ui.custom() factory signature changed from (tui, theme,

破坏性变更7

  • ctx.ui.custom() factory signature changed from (tui, theme, done) to (tui, theme, keybindings, done) for keybinding access in custom components
  • LoadedExtension type renamed to Extension
  • LoadExtensionsResult.setUIContext() removed, replaced with runtime: ExtensionRuntime
  • ExtensionRunner constructor now requires runtime: ExtensionRuntime as second parameter
  • ExtensionRunner.initialize() signature changed from options object to positional params (actions, contextActions, commandContextActions?, uiContext?)
  • ExtensionRunner.getHasUI() renamed to hasUI()
  • OpenAI Codex model aliases removed (gpt-5, gpt-5-mini, gpt-5-nano, codex-mini-latest). Use canonical IDs: gpt-5.1, gpt-5.1-codex-mini, gpt-5.2, gpt-5.2-codex. (#536 by @ghoulr)

新增8

  • --no-extensions flag to disable extension discovery while still allowing explicit -e paths (#524 by @cv)
  • SDK: InteractiveMode, runPrintMode(), runRpcMode() exported for building custom run modes. See docs/sdk.md.
  • PI_SKIP_VERSION_CHECK environment variable to disable new version notifications at startup (#549 by @aos)
  • thinkingBudgets setting to customize token budgets per thinking level for token-based providers (#529 by @melihmucuk)
  • Extension UI dialogs (ctx.ui.select(), ctx.ui.confirm(), ctx.ui.input()) now support a timeout option with live countdown display (#522 by @nicobailon)
  • Extensions can now provide custom editor components via ctx.ui.setEditorComponent(). See examples/extensions/modal-editor.ts and docs/tui.md Pattern 7.
  • Extension factories can now be async, enabling dynamic imports and lazy-loaded dependencies (#513 by @austinm911)
  • ctx.shutdown() is now available in extension contexts for requesting a graceful shutdown. In interactive mode, shutdown is deferred until the agent becomes idle (after processing all queued steering and follow-up messages). In RPC mode, shutdown is deferred until after completing the current command response. In print mode, shutdown is a no-op as the process exits automatically when prompts complete. (#542 by @kaofelix)

修复7

  • Default thinking level from settings now applies correctly when enabledModels is configured (#540 by @ferologics)
  • External edits to settings.json while pi is running are now preserved when pi saves settings (#527 by @ferologics)
  • Overflow-based compaction now skips if error came from a different model or was already handled by a previous compaction (#535 by @mitsuhiko)
  • OpenAI Codex context window reduced from 400k to 272k tokens to match Codex CLI defaults and prevent 400 errors (#536 by @ghoulr)
  • Context overflow detection now recognizes context_length_exceeded errors.
  • Key presses no longer dropped when input is batched over SSH (#538)
  • Clipboard image support now works on Alpine Linux and other musl-based distros (#533)
0.37.8 2026-01-07 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.37.7 2026-01-07 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.37.6 2026-01-06 新增 2 Extension UI dialogs (ctx.ui.select(), ctx.ui.confirm(), ctx

新增2

  • Extension UI dialogs (ctx.ui.select(), ctx.ui.confirm(), ctx.ui.input()) now accept an optional AbortSignal to programmatically dismiss dialogs. Useful for implementing timeouts. See examples/extensions/timed-confirm.ts. (#474)
  • HTML export now shows bridge prompts in model change messages for Codex sessions (#510 by @mitsuhiko)
0.37.5 2026-01-06 新增 10 ExtensionAPI: setModel(), getThinkingLevel(), setThinkingLev

新增10

  • ExtensionAPI: setModel(), getThinkingLevel(), setThinkingLevel() methods for extensions to change model and thinking level at runtime (#509)
  • Exported truncation utilities for custom tools: truncateHead, truncateTail, truncateLine, formatSize, DEFAULT_MAX_BYTES, DEFAULT_MAX_LINES, TruncationOptions, TruncationResult
  • New example truncated-tool.ts demonstrating proper output truncation with custom rendering for extensions
  • New example preset.ts demonstrating preset configurations with model/thinking/tools switching (#347)
  • Documentation for output truncation best practices in docs/extensions.md
  • Exported all UI components for extensions: ArminComponent, AssistantMessageComponent, BashExecutionComponent, BorderedLoader, BranchSummaryMessageComponent, CompactionSummaryMessageComponent, CustomEditor, CustomMessageComponent, DynamicBorder, ExtensionEditorComponent, ExtensionInputComponent, ExtensionSelectorComponent, FooterComponent, LoginDialogComponent, ModelSelectorComponent, OAuthSelectorComponent, SessionSelectorComponent, SettingsSelectorComponent, ShowImagesSelectorComponent, ThemeSelectorComponent, ThinkingSelectorComponent, ToolExecutionComponent, TreeSelectorComponent, UserMessageComponent, UserMessageSelectorComponent, plus utilities renderDiff, truncateToVisualLines
  • docs/tui.md: Common Patterns section with copy-paste code for SelectList, BorderedLoader, SettingsList, setStatus, setWidget, setFooter
  • docs/tui.md: Key Rules section documenting critical patterns for extension UI development
  • docs/extensions.md: Exhaustive example links for all ExtensionAPI methods and events
  • System prompt now references docs/tui.md for TUI component development
0.37.4 2026-01-06 新增 2变更 2修复 2 Session picker (pi -r) and --session flag now support search

新增2

  • Session picker (pi -r) and --session flag now support searching/resuming by session ID (UUID prefix) (#495 by @arunsathiya)
  • Extensions can now replace the startup header with ctx.ui.setHeader(), see examples/extensions/custom-header.ts (#500 by @tudoroancea)

变更2

  • Startup help text: fixed misleading "ctrl+k to delete line" to "ctrl+k to delete to end"
  • Startup help text and /hotkeys: added !! shortcut for running bash without adding output to context

修复2

  • Queued steering/follow-up messages no longer wipe unsent editor input (#503 by @tmustier)
  • OAuth token refresh failure no longer crashes app at startup, allowing user to /login to re-authenticate (#498)
0.37.3 2026-01-06 新增 4修复 3 Extensions can now replace the footer with ctx.ui.setFooter(

新增4

  • Extensions can now replace the footer with ctx.ui.setFooter(), see examples/extensions/custom-footer.ts (#481)
  • Session ID is now forwarded to LLM providers for session-based caching (used by OpenAI Codex for prompt caching).
  • Added blockImages setting to prevent images from being sent to LLM providers (#492 by @jsinge97)
  • Extensions can now send user messages via pi.sendUserMessage() (#483)

修复3

  • Add minimatch as a direct dependency for explicit imports.
  • Status bar now shows correct git branch when running in a git worktree (#490 by @kcosr)
  • Interactive mode: Ctrl+V clipboard image paste now works on Wayland sessions by using wl-paste with xclip fallback (#488 by @ghoulr)
0.37.2 2026-01-05 修复 3 Extension directories in settings.json now respect package.j

修复3

  • Extension directories in settings.json now respect package.json manifests, matching global extension behavior (#480 by @prateekmedia)
  • Share viewer: deep links now scroll to the target message when opened via /share
  • Bash tool now handles spawn errors gracefully instead of crashing the agent (missing cwd, invalid shell path) (#479 by @robinwander)
0.37.1 2026-01-05 修复 1 Share viewer: copy-link buttons now generate correct URLs wh

修复1

  • Share viewer: copy-link buttons now generate correct URLs when session is viewed via /share (iframe context)
0.37.0 2026-01-05 新增 3变更 3修复 7 Share viewer: copy-link button on messages to share URLs tha

新增3

  • Share viewer: copy-link button on messages to share URLs that navigate directly to a specific message (#477 by @lockmeister)
  • Extension example: add claude-rules to load .claude/rules/ entries into the system prompt (#461 by @vaayne)
  • Headless OAuth login: all providers now show paste input for manual URL/code entry, works over SSH without DISPLAY (#428 by @ben-vargas, #468 by @crcatala)

变更3

  • OAuth login UI now uses dedicated dialog component with consistent borders
  • Assume truecolor support for all terminals except dumb, empty, or linux (fixes colors over SSH)
  • OpenAI Codex clean-up: removed per-thinking-level model variants, thinking level is now set separately and the provider clamps to what each model supports internally (initial implementation in #472 by @ben-vargas)

修复7

  • Messages submitted during compaction are queued and delivered after compaction completes, preserving steering and follow-up behavior. Extension commands execute immediately during compaction. (#476 by @tmustier)
  • Managed binaries (fd, rg) now stored in ~/.pi/agent/bin/ instead of tools/, eliminating false deprecation warnings (#470 by @mcinteerj)
  • Extensions defined in settings.json were not loaded (#463 by @melihmucuk)
  • OAuth refresh no longer logs users out when multiple pi instances are running (#466 by @Cursivez)
  • Migration warnings now ignore fd.exe and rg.exe in tools/ on Windows (#458 by @carlosgtrz)
  • CI: add examples/extensions/with-deps to workspaces to fix typecheck (#467 by @aliou)
  • SDK: passing extensions: [] now disables extension discovery as documented (#465 by @aliou)
0.36.0 2026-01-05 新增 1 Experimental: OpenAI Codex OAuth provider support: access Co

新增1

  • Experimental: OpenAI Codex OAuth provider support: access Codex models via ChatGPT Plus/Pro subscription using /login openai-codex (#451 by @kim0)
0.35.0 2026-01-05 其他 3其他 3其他 31其他 9其他 16其他 1破坏性变更 5变更 7 docs/extensions.md - Full API reference

未分组3

  • docs/extensions.md - Full API reference
  • README.md - Extensions section with examples
  • examples/extensions/ - Working examples

未分组3

  • docs/extensions.md - Full API reference
  • README.md - Extensions section with examples
  • examples/extensions/ - Working examples

Extensions Migration31

  • commands/ directories are automatically renamed to prompts/ on startup (both ~/.pi/agent/commands/ and .pi/commands/)

    "name": "my-extension-package",

    "dependencies": { "zod": "^3.0.0" },

    "pi": {

    "extensions": ["./src/main.ts", "./src/tools.ts"]

    }

    }

    No recursion beyond one level. Complex packages must use the `package.json` manifest. Dependencies are resolved via jiti, and extensions can be published to and installed from npm.
**Type renames:**
  • HookAPI → ExtensionAPI
  • HookContext → ExtensionContext
  • HookCommandContext → ExtensionCommandContext
  • HookUIContext → ExtensionUIContext
  • CustomToolAPI → ExtensionAPI (merged)
  • CustomToolContext → ExtensionContext (merged)
  • CustomToolUIContext → ExtensionUIContext
  • CustomTool → ToolDefinition
  • CustomToolFactory → ExtensionFactory
  • HookMessage → CustomMessage

    pi.on("tool_call", async (event, ctx) => { ... });

    pi.registerTool({ name: "my_tool", ... });

    }

    **Custom tools now have full context access.** Tools registered via `pi.registerTool()` now receive the same `ctx` object that event handlers receive. Previously, custom tools had limited context. Now all extension code shares the same capabilities:
  • pi.registerTool() - Register tools the LLM can call
  • pi.registerCommand() - Register commands like /mycommand
  • pi.registerShortcut() - Register keyboard shortcuts (shown in /hotkeys)
  • pi.registerFlag() - Register CLI flags (shown in --help)
  • pi.registerMessageRenderer() - Custom TUI rendering for message types
  • pi.on() - Subscribe to lifecycle events (tool_call, session_start, etc.)
  • pi.sendMessage() - Inject messages into the conversation
  • pi.appendEntry() - Persist custom data in session (survives restart/branch)
  • pi.exec() - Run shell commands
  • pi.getActiveTools() / pi.setActiveTools() - Dynamic tool enable/disable
  • pi.getAllTools() - List all available tools
  • pi.events - Event bus for cross-extension communication
  • ctx.ui.confirm() / select() / input() - User prompts
  • ctx.ui.notify() - Toast notifications
  • ctx.ui.setStatus() - Persistent status in footer (multiple extensions can set their own)
  • ctx.ui.setWidget() - Widget display above editor
  • ctx.ui.setTitle() - Set terminal window title
  • ctx.ui.custom() - Full TUI component with keyboard handling
  • ctx.ui.editor() - Multi-line text editor with external editor support
  • ctx.sessionManager - Read session entries, get branch history

    "hooks": ["./my-hook.ts"],

    "customTools": ["./my-tool.ts"]

    }

    // After

    {

    "extensions": ["./my-extension.ts"]

    }

    **CLI changes:**

    # Before

    pi --hook ./safety.ts --tool ./todo.ts

    # After

    pi --extension ./safety.ts -e ./todo.ts

Prompt Templates Migration9

  • FileSlashCommand → PromptTemplate
  • LoadSlashCommandsOptions → LoadPromptTemplatesOptions
  • discoverSlashCommands() → discoverPromptTemplates()
  • loadSlashCommands() → loadPromptTemplates()
  • expandSlashCommand() → expandPromptTemplate()
  • getCommandsDir() → getPromptsDir()
  • CreateAgentSessionOptions.slashCommands → .promptTemplates
  • AgentSession.fileCommands → .promptTemplates
  • PromptOptions.expandSlashCommands → .expandPromptTemplates

SDK Migration16

  • discoverAndLoadHooks() → discoverAndLoadExtensions()
  • discoverAndLoadCustomTools() → merged into discoverAndLoadExtensions()
  • loadHooks() → loadExtensions()
  • loadCustomTools() → merged into loadExtensions()
  • HookRunner → ExtensionRunner
  • wrapToolsWithHooks() → wrapToolsWithExtensions()
  • wrapToolWithHooks() → wrapToolWithExtensions()
  • .hooks → removed (use .additionalExtensionPaths for paths)
  • .additionalHookPaths → .additionalExtensionPaths
  • .preloadedHooks → .preloadedExtensions
  • .customTools type changed: Array<{ path?; tool: CustomTool }> → ToolDefinition[]
  • .additionalCustomToolPaths → merged into .additionalExtensionPaths
  • .slashCommands → .promptTemplates
  • .hookRunner → .extensionRunner
  • .fileCommands → .promptTemplates
  • .sendHookMessage() → .sendCustomMessage()

Session Migration1

  • Message role "hookMessage" → "custom"

破坏性变更5

  • Settings: hooks and customTools arrays replaced with single extensions array
  • CLI: --hook and --tool flags replaced with --extension / -e
  • Directories: hooks/, tools/ → extensions/; commands/ → prompts/
  • Types: See type renames above
  • SDK: See SDK migration above

变更7

  • Extensions can have their own package.json with dependencies (resolved via jiti)
  • Documentation: docs/hooks.md and docs/custom-tools.md merged into docs/extensions.md
  • Examples: examples/hooks/ and examples/custom-tools/ merged into examples/extensions/
  • README: Extensions section expanded with custom tools, commands, events, state persistence, shortcuts, flags, and UI examples
  • SDK: customTools option now accepts ToolDefinition[] directly (simplified from Array<{ path?, tool }>)
  • SDK: extensions option accepts ExtensionFactory[] for inline extensions
  • SDK: additionalExtensionPaths replaces both additionalHookPaths and additionalCustomToolPaths
0.34.2 2026-01-04 该版本在 CHANGELOG 里只有版本号与日期，没记录条目

该版本只打了个版本号（CHANGELOG 里空条目），通常是发版顺序调整或空提交。

0.34.1 2026-01-04 新增 1变更 1 Hook API: ctx.ui.setTitle(title) allows hooks to set the ter

新增1

  • Hook API: ctx.ui.setTitle(title) allows hooks to set the terminal window/tab title (#446 by @aliou)

变更1

  • Expanded keybinding documentation to list all 32 supported symbol keys with notes on ctrl+symbol behavior (#450 by @kaofelix)
0.34.0 2026-01-04 新增 18变更 1修复 2 Hook API: pi.getActiveTools() and pi.setActiveTools(toolName

新增18

  • Hook API: pi.getActiveTools() and pi.setActiveTools(toolNames) for dynamically enabling/disabling tools from hooks
  • Hook API: pi.getAllTools() to enumerate all configured tools (built-in via --tools or default, plus custom tools)
  • Hook API: pi.registerFlag(name, options) and pi.getFlag(name) for hooks to register custom CLI flags (parsed automatically)
  • Hook API: pi.registerShortcut(shortcut, options) for hooks to register custom keyboard shortcuts using KeyId (e.g., Key.shift("p")). Conflicts with built-in shortcuts are skipped, conflicts between hooks logged as warnings.
  • Hook API: ctx.ui.setWidget(key, content) for status displays above the editor. Accepts either a string array or a component factory function.
  • Hook API: theme.strikethrough(text) for strikethrough text styling
  • Hook API: before_agent_start handlers can now return systemPromptAppend to dynamically append text to the system prompt for that turn. Multiple hooks' appends are concatenated.
  • Hook API: before_agent_start handlers can now return multiple messages (all are injected, not just the first)
  • /hotkeys command now shows hook-registered shortcuts in a separate "Hooks" section
  • New example hook: plan-mode.ts - Claude Code-style read-only exploration mode:

    · Toggle via /plan command, Shift+P shortcut, or --plan CLI flag

    · Read-only tools: read, bash, grep, find, ls (no edit/write)

    · Bash commands restricted to non-destructive operations (blocks rm, mv, git commit, npm install, etc.)

    · Interactive prompt after each response: execute plan, stay in plan mode, or refine

    · Todo list widget showing progress with checkboxes and strikethrough for completed items

    · Each todo has a unique ID; agent marks items done by outputting [DONE:id]

    · Progress updates via agent_end hook (parses completed items from final message)

    · /todos command to view current plan progress

    · Shows ⏸ plan indicator in footer when in plan mode, 📋 2/5 when executing

    · State persists across sessions (including todo progress)

  • New example hook: tools.ts - Interactive /tools command to enable/disable tools with session persistence
  • New example hook: pirate.ts - Demonstrates systemPromptAppend to make the agent speak like a pirate
  • Tool registry now contains all built-in tools (read, bash, edit, write, grep, find, ls) even when --tools limits the initially active set. Hooks can enable any tool from the registry via pi.setActiveTools().
  • System prompt now automatically rebuilds when tools change via setActiveTools(), updating tool descriptions and guidelines to match the new tool set
  • Hook errors now display full stack traces for easier debugging
  • Event bus (pi.events) for tool/hook communication: shared pub/sub between custom tools and hooks
  • Custom tools now have pi.sendMessage() to send messages directly to the agent session without needing the event bus
  • sendMessage() supports deliverAs: "nextTurn" to queue messages for the next user prompt

变更1

  • Removed image placeholders after copy & paste, replaced with inserting image file paths directly. (#442 by @mitsuhiko)

修复2

  • Fixed potential text decoding issues in bash executor by using streaming TextDecoder instead of Buffer.toString()
  • External editor (Ctrl-G) now shows full pasted content instead of [paste #N ...] placeholders (#444 by @aliou)
0.33.0 2026-01-04 破坏性变更 1新增 3修复 1 Key detection functions removed from @mariozechner/pi-tui

破坏性变更1

  • Key detection functions removed from @mariozechner/pi-tui: All isXxx() key detection functions (isEnter(), isEscape(), isCtrlC(), etc.) have been removed. Use matchesKey(data, keyId) instead (e.g., matchesKey(data, "enter"), matchesKey(data, "ctrl+c")). This affects hooks and custom tools that use ctx.ui.custom() with keyboard input handling. (#405)

新增3

  • Clipboard image paste support via Ctrl+V. Images are saved to a temp file and attached to the message. Works on macOS, Windows, and Linux (X11). (#419)
  • Configurable keybindings via ~/.pi/agent/keybindings.json. All keyboard shortcuts (editor navigation, deletion, app actions like model cycling, etc.) can now be customized. Supports multiple bindings per action. (#405 by @hjanuschka)
  • /quit and /exit slash commands to gracefully exit the application. Unlike double Ctrl+C, these properly await hook and custom tool cleanup handlers before exiting. (#426 by @ben-vargas)

修复1

  • Subagent example README referenced incorrect filename subagent.ts instead of index.ts (#427 by @Whamp)
0.32.3 2026-01-03 修复 4 --list-models no longer shows Google Vertex AI models withou

修复4

  • --list-models no longer shows Google Vertex AI models without explicit authentication configured
  • JPEG/GIF/WebP images not displaying in terminals using Kitty graphics protocol (Kitty, Ghostty, WezTerm). The protocol requires PNG format, so non-PNG images are now converted before display.
  • Version check URL typo preventing update notifications from working (#423 by @skuridin)
  • Large images exceeding Anthropic's 5MB limit now retry with progressive quality/size reduction (#424 by @mitsuhiko)
0.32.2 2026-01-03 新增 1变更 1修复 1 $ARGUMENTS syntax for custom slash commands as alternative t

新增1

  • $ARGUMENTS syntax for custom slash commands as alternative to $@ for all arguments joined. Aligns with patterns used by Claude, Codex, and OpenCode. Both syntaxes remain fully supported. (#418 by @skuridin)

变更1

  • Slash commands and hook commands now work during streaming: Previously, using a slash command or hook command while the agent was streaming would crash with "Agent is already processing". Now:

    · Hook commands execute immediately (they manage their own LLM interaction via pi.sendMessage())

    · File-based slash commands are expanded and queued via steer/followUp

    · steer() and followUp() now expand file-based slash commands and error on hook commands (hook commands cannot be queued)

    · prompt() accepts new streamingBehavior option ("steer" or "followUp") to specify queueing behavior during streaming

    · RPC prompt command now accepts optional streamingBehavior field

    (#420)

修复1

  • Slash command argument substitution now processes positional arguments ($1, $2, etc.) before all-arguments ($@, $ARGUMENTS) to prevent recursive substitution when argument values contain dollar-digit patterns like $100. (#418 by @skuridin)
0.32.1 2026-01-03 新增 1修复 1 Shell commands without context contribution: use !!command t

新增1

  • Shell commands without context contribution: use !!command to execute a bash command that is shown in the TUI and saved to session history but excluded from LLM context. Useful for running commands you don't want the AI to see. (#414)

修复1

  • Edit tool diff not displaying in TUI due to race condition between async preview computation and tool execution
0.32.0 2026-01-03 破坏性变更 6新增 7变更 1修复 4 Queue API replaced with steer/followUp

破坏性变更6

  • Queue API replaced with steer/followUp: The queueMessage() method has been split into two methods with different delivery semantics (#403):

    · steer(text): Interrupts the agent mid-run (Enter while streaming). Delivered after current tool execution.

    · followUp(text): Waits until the agent finishes (Alt+Enter while streaming). Delivered only when agent stops.

  • Settings renamed: queueMode setting renamed to steeringMode. Added new followUpMode setting. Old settings.json files are migrated automatically.
  • AgentSession methods renamed:

    · queueMessage() → steer() and followUp()

    · queueMode getter → steeringMode and followUpMode getters

    · setQueueMode() → setSteeringMode() and setFollowUpMode()

    · queuedMessageCount → pendingMessageCount

    · getQueuedMessages() → getSteeringMessages() and getFollowUpMessages()

    · clearQueue() now returns { steering: string[], followUp: string[] }

    · hasQueuedMessages() → hasPendingMessages()

  • Hook API signature changed: pi.sendMessage() second parameter changed from triggerTurn?: boolean to options?: { triggerTurn?, deliverAs? }. Use deliverAs: "followUp" for follow-up delivery. Affects both hooks and internal sendHookMessage() method.
  • RPC API changes:

    · queue_message command → steer and follow_up commands

    · set_queue_mode command → set_steering_mode and set_follow_up_mode commands

    · RpcSessionState.queueMode → steeringMode and followUpMode

  • Settings UI: "Queue mode" setting split into "Steering mode" and "Follow-up mode"

新增7

  • Configurable double-escape action: choose whether double-escape with empty editor opens /tree (default) or /branch. Configure via /settings or doubleEscapeAction in settings.json (#404)
  • Vertex AI provider (google-vertex): access Gemini models via Google Cloud Vertex AI using Application Default Credentials (#300 by @default-anton)
  • Built-in provider overrides in models.json: override just baseUrl to route a built-in provider through a proxy while keeping all its models, or define models to fully replace the provider (#406 by @yevhen)
  • Automatic image resizing: images larger than 2000x2000 are resized for better model compatibility. Original dimensions are injected into the prompt. Controlled via /settings or images.autoResize in settings.json. (#402 by @mitsuhiko)
  • Alt+Enter keybind to queue follow-up messages while agent is streaming
  • Theme and ThemeColor types now exported for hooks using ctx.ui.custom()
  • Terminal window title now displays "pi - dirname" to identify which project session you're in (#407 by @kaofelix)

变更1

  • Editor component now uses word wrapping instead of character-level wrapping for better readability (#382 by @nickseelert)

修复4

  • /model selector now opens instantly instead of waiting for OAuth token refresh. Token refresh is deferred until a model is actually used.
  • Shift+Space, Shift+Backspace, and Shift+Delete now work correctly in Kitty-protocol terminals (Kitty, WezTerm, etc.) instead of being silently ignored (#411 by @nathyong)
  • AgentSession.prompt() now throws if called while the agent is already streaming, preventing race conditions. Use steer() or followUp() to queue messages during streaming.
  • Ctrl+C now works like Escape in selector components, so mashing Ctrl+C will eventually close the program (#400 by @mitsuhiko)
0.31.1 2026-01-02 修复 2 Model selector no longer allows negative index when pressing

修复2

  • Model selector no longer allows negative index when pressing arrow keys before models finish loading (#398 by @mitsuhiko)
  • Type guard functions (isBashToolResult, etc.) now exported at runtime, not just in type declarations (#397)
0.31.0 2026-01-02 其他 26其他 11其他 24其他 8其他 3其他 12新增 9变更 3修复 16 HookEventContext → HookContext

Hooks Migration26

  • HookEventContext → HookContext
  • HookCommandContext is now a new interface extending HookContext with session control methods
  • The monolithic session event is now split into granular events: session_start, session_before_switch, session_switch, session_before_branch, session_branch, session_before_compact, session_compact, session_shutdown
  • session_before_switch and session_switch events now include reason: "new" | "resume" to distinguish between /new and /resume
  • New session_before_tree and session_tree events for /tree navigation (hook can provide custom branch summary)
  • New before_agent_start event: inject messages before the agent loop starts
  • New context event: modify messages non-destructively before each LLM call
  • Session entries are no longer passed in events. Use ctx.sessionManager.getEntries() or ctx.sessionManager.getBranch() instead
  • pi.send(text, attachments?) → pi.sendMessage(message, triggerTurn?) (creates CustomMessageEntry)
  • New pi.appendEntry(customType, data?) for hook state persistence (not in LLM context)
  • New pi.registerCommand(name, options) for custom slash commands (handler receives HookCommandContext)
  • New pi.registerMessageRenderer(customType, renderer) for custom TUI rendering
  • New ctx.isIdle(), ctx.abort(), ctx.hasQueuedMessages() for agent state (available in all events)
  • New ctx.ui.editor(title, prefill?) for multi-line text editing with Ctrl+G external editor support
  • New ctx.ui.custom(component) for full TUI component rendering with keyboard focus
  • New ctx.ui.setStatus(key, text) for persistent status text in footer (multiple hooks can set their own)
  • New ctx.ui.theme getter for styling text with theme colors
  • ctx.exec() moved to pi.exec()
  • ctx.sessionFile → ctx.sessionManager.getSessionFile()
  • New ctx.modelRegistry and ctx.model for API key resolution
  • ctx.waitForIdle() - wait for agent to finish streaming
  • ctx.newSession(options?) - create new sessions with optional setup callback
  • `ctx.fork(entryId) - fork from a specific entry, creating a new session file
  • ctx.navigateTree(targetId, options?) - navigate the session tree
  • hookTimeout setting (hooks no longer have timeouts; use Ctrl+C to abort)
  • resolveApiKey parameter (use ctx.modelRegistry.getApiKey(model))

Custom Tools Migration11

  • CustomAgentTool → CustomTool
  • ToolAPI → CustomToolAPI
  • ToolContext → CustomToolContext
  • ToolSessionEvent → CustomToolSessionEvent
  • ctx.isIdle() - check if agent is streaming
  • ctx.hasQueuedMessages() - check if user has queued messages (skip interactive prompts)
  • ctx.abort() - abort current operation (fire-and-forget)
  • CustomToolSessionEvent now only has reason and previousSessionFile
  • Session entries are no longer in the event. Use ctx.sessionManager.getBranch() or ctx.sessionManager.getEntries() to reconstruct state
  • Reasons: "start" | "switch" | "branch" | "tree" | "shutdown" (no separate "new" reason; /new triggers "switch")
  • dispose() method removed. Use onSession with reason: "shutdown" for cleanup

SDK Migration24

  • CustomAgentTool → CustomTool
  • AppMessage → AgentMessage
  • sessionFile returns string | undefined (was string | null)
  • model returns Model | undefined (was Model | null)
  • Attachment type removed. Use ImageContent from @mariozechner/pi-ai instead. Add images directly to message content arrays.
  • branch(entryIndex: number) → branch(entryId: string)
  • getUserMessagesForBranching() returns { entryId, text } instead of { entryIndex, text }
  • reset() → newSession(options?) where options has optional parentSession for lineage tracking
  • newSession() and switchSession() now return Promise<boolean> (false if cancelled by hook)
  • New navigateTree(targetId, options?) for in-place tree navigation
  • New sendHookMessage(message, triggerTurn?) for hook message injection
  • Method renames: saveXXX() → appendXXX() (e.g., appendMessage, appendCompaction)
  • branchInPlace() → branch()
  • reset() → newSession(options?) with optional parentSession for lineage tracking
  • createBranchedSessionFromEntries(entries, index) → createBranchedSession(leafId)
  • SessionHeader.branchedFrom → SessionHeader.parentSession
  • saveCompaction(entry) → appendCompaction(summary, firstKeptEntryId, tokensBefore, details?)
  • getEntries() now excludes the session header (use getHeader() separately)
  • getSessionFile() returns string | undefined (undefined for in-memory sessions)
  • New tree methods: getTree(), getBranch(), getLeafId(), getLeafEntry(), getEntry(), getChildren(), getLabel()
  • New append methods: appendCustomEntry(), appendCustomMessageEntry(), appendLabelChange()
  • New branch methods: branch(entryId), branchWithSummary()

    discoverAuthStorage,

    discoverModels,

    } from "@mariozechner/pi-coding-agent";

    const authStorage = discoverAuthStorage(); // ~/.pi/agent/auth.json

    const modelRegistry = discoverModels(authStorage); // + ~/.pi/agent/models.json

    // Get all models (built-in + custom)

    const allModels = modelRegistry.getAll();

    // Get only models with valid API keys

    const available = await modelRegistry.getAvailable();

    // Find specific model

    const model = modelRegistry.find("anthropic", "claude-sonnet-4-20250514");

    // Get API key for a model

    const apiKey = await modelRegistry.getApiKey(model);

    This replaces the old `resolveApiKey` callback pattern. Hooks and custom tools access it via `ctx.modelRegistry`.
**Renamed exports:**
  • messageTransformer → convertToLlm
  • SessionContext alias LoadedSession removed

RPC Migration8

  • reset command → new_session command with optional parentSession field
  • branch command: entryIndex → entryId
  • get_branch_messages response: entryIndex → entryId
  • Messages are now AgentMessage (was AppMessage)
  • prompt command: attachments field replaced with images field using ImageContent format
  • auto_compaction_start now includes reason field ("threshold" or "overflow")
  • auto_compaction_end now includes willRetry field
  • compact response includes full CompactionResult (summary, firstKeptEntryId, tokensBefore, details)

Structured Compaction3

  • Clear sections: Goal, Progress, Key Information, File Operations
  • File tracking: readFiles and modifiedFiles arrays in details, accumulated across compactions
  • Conversations are serialized to text before summarization to prevent the model from "continuing" them

Interactive Mode12

  • Navigate the full session tree in-place
  • Search by typing, page with ←/→
  • Filter modes (Ctrl+O): default → no-tools → user-only → labeled-only → all
  • Press l to label entries as bookmarks
  • Selecting a branch switches context and optionally injects a summary of the abandoned branch
  • Bookmark any entry via /tree → select → l
  • Labels appear in tree view and persist as LabelEntry
  • selectedBg: background for selected/highlighted items in tree selector and other components
  • customMessageBg: background for hook-injected messages (CustomMessageEntry)
  • customMessageText: text color for hook messages
  • customMessageLabel: label color for hook messages (the [customType] prefix)
  • enabledModels: allowlist models in settings.json (same format as --models CLI)

新增9

  • ctx.ui.setStatus(key, text) for hooks to display persistent status text in the footer (#385 by @prateekmedia)
  • ctx.ui.theme getter for styling status text and other output with theme colors
  • /share command to upload session as a secret GitHub gist and get a shareable URL via pi.dev (#380)
  • HTML export now includes a tree visualization sidebar for navigating session branches (#375)
  • HTML export supports keyboard shortcuts: Ctrl+T to toggle thinking blocks, Ctrl+O to toggle tool outputs
  • HTML export supports theme-configurable background colors via optional export section in theme JSON (#387 by @mitsuhiko)
  • HTML export syntax highlighting now uses theme colors and matches TUI rendering
  • Snake game example hook: Demonstrates ui.custom(), registerCommand(), and session persistence. See examples/hooks/snake.ts.
  • thinkingText theme token: Configurable color for thinking block text. (#366 by @paulbettner)

变更3

  • Entry IDs: Session entries now use short 8-character hex IDs instead of full UUIDs
  • API key priority: ANTHROPIC_OAUTH_TOKEN now takes precedence over ANTHROPIC_API_KEY
  • HTML export template split into separate files (template.html, template.css, template.js) for easier maintenance

修复16

  • HTML export now properly sanitizes user messages containing HTML tags like <style> that could break DOM rendering
  • Crash when displaying bash output containing Unicode format characters like U+0600-U+0604 (#372 by @HACKE-RC)
  • Footer shows full session stats: Token usage and cost now include all messages, not just those after compaction. (#322)
  • Status messages spam chat log: Rapidly changing settings (e.g., thinking level via Shift+Tab) would add multiple status lines. Sequential status updates now coalesce into a single line. (#365 by @paulbettner)
  • Toggling thinking blocks during streaming shows nothing: Pressing Ctrl+T while streaming would hide the current message until streaming completed.
  • Resuming session resets thinking level to off: Initial model and thinking level were not saved to session file, causing --resume/--continue to default to off. (#342 by @aliou)
  • Hook tool_result event ignores errors from custom tools: The tool_result hook event was never emitted when tools threw errors, and always had isError: false for successful executions. Now emits the event with correct isError value in both success and error cases. (#374 by @nicobailon)
  • Edit tool fails on Windows due to CRLF line endings: Files with CRLF line endings now match correctly when LLMs send LF-only text. Line endings are normalized before matching and restored to original style on write. (#355 by @Pratham-Dubey)
  • Edit tool fails on files with UTF-8 BOM: Files with UTF-8 BOM marker could cause "text not found" errors since the LLM doesn't include the invisible BOM character. BOM is now stripped before matching and restored on write. (#394 by @prathamdby)
  • Use bash instead of sh on Unix: Fixed shell commands using /bin/sh instead of /bin/bash on Unix systems. (#328 by @dnouri)
  • OAuth login URL clickable: Made OAuth login URLs clickable in terminal. (#349 by @Cursivez)
  • Improved error messages: Better error messages when apiKey or model are missing. (#346 by @ronyrus)
  • Session file validation: findMostRecentSession() now validates session headers before returning, preventing non-session JSONL files from being loaded
  • Compaction error handling: generateSummary() and generateTurnPrefixSummary() now throw on LLM errors instead of returning empty strings
  • Compaction with branched sessions: Fixed compaction incorrectly including entries from abandoned branches, causing token overflow errors. Compaction now uses sessionManager.getPath() to work only on the current branch path, eliminating 80+ lines of duplicate entry collection logic between prepareCompaction() and compact()
  • enabledModels glob patterns: --models and enabledModels now support glob patterns like github-copilot/* or *sonnet*. Previously, patterns were only matched literally or via substring search. (#337)
2025-1275 个版本
0.30.2 2025-12-26 变更 1 Consolidated migrations

变更1

  • Consolidated migrations: Moved auth migration from AuthStorage.migrateLegacy() to new migrations.ts module.
0.30.1 2025-12-26 修复 2 Sessions saved to wrong directory

修复2

  • Sessions saved to wrong directory: In v0.30.0, sessions were being saved to ~/.pi/agent/ instead of ~/.pi/agent/sessions/<encoded-cwd>/, breaking --resume and /resume. Misplaced sessions are automatically migrated on startup. (#320 by @aliou)
  • Custom system prompts missing context: When using a custom system prompt string, project context files (AGENTS.md), skills, date/time, and working directory were not appended. (#321)
0.30.0 2025-12-25 破坏性变更 1新增 2 SessionManager API

破坏性变更1

  • SessionManager API: The second parameter of create(), continueRecent(), and list() changed from agentDir to sessionDir. When provided, it specifies the session directory directly (no cwd encoding). When omitted, uses default (~/.pi/agent/sessions/<encoded-cwd>/). open() no longer takes agentDir. (#313)

新增2

  • --session-dir flag: Use a custom directory for sessions instead of the default ~/.pi/agent/sessions/<encoded-cwd>/. Works with -c (continue) and -r (resume) flags. (#313 by @scutifer)
  • Reverse model cycling and model selector: Shift+Ctrl+P cycles models backward, Ctrl+L opens model selector (retaining text in editor). (#315 by @mitsuhiko)
0.29.1 2025-12-25 新增 2修复 1 Automatic custom system prompt loading

新增2

  • Automatic custom system prompt loading: Pi now auto-loads SYSTEM.md files to replace the default system prompt. Project-local .pi/SYSTEM.md takes precedence over global ~/.pi/agent/SYSTEM.md. CLI --system-prompt flag overrides both. (#309)
  • Unified /settings command: New settings menu consolidating thinking level, theme, queue mode, auto-compact, show images, hide thinking, and collapse changelog. Replaces individual /thinking, /queue, /theme, /autocompact, and /show-images commands. (#310)

修复1

  • Custom tools/hooks with typebox subpath imports: Fixed jiti alias for @sinclair/typebox to point to package root instead of entry file, allowing imports like @sinclair/typebox/compiler to resolve correctly. (#311 by @kim0)
0.29.0 2025-12-25 破坏性变更 1新增 3修复 1 Renamed /clear to /new

破坏性变更1

  • Renamed /clear to /new: The command to start a fresh session is now /new. Hook event reasons before_clear/clear are now before_new/new. Merry Christmas @mitsuhiko! (#305)

新增3

  • Auto-space before pasted file paths: When pasting a file path (starting with /, ~, or .) after a word character, a space is automatically prepended. (#307 by @mitsuhiko)
  • Word navigation in input fields: Added Ctrl+Left/Right and Alt+Left/Right for word-by-word cursor movement. (#306 by @kim0)
  • Full Unicode input: Input fields now accept Unicode characters beyond ASCII. (#306 by @kim0)

修复1

  • Readline-style Ctrl+W: Now skips trailing whitespace before deleting the preceding word, matching standard readline behavior. (#306 by @kim0)
0.28.0 2025-12-25 变更 3修复 1 Credential storage refactored

变更3

  • Credential storage refactored: API keys and OAuth tokens are now stored in ~/.pi/agent/auth.json instead of oauth.json and settings.json. Existing credentials are automatically migrated on first run. (#296)
  • SDK API changes (#296):

    · Added AuthStorage class for credential management (API keys and OAuth tokens)

    · Added ModelRegistry class for model discovery and API key resolution

    · Added discoverAuthStorage() and discoverModels() discovery functions

    · createAgentSession() now accepts authStorage and modelRegistry options

    · Removed configureOAuthStorage(), defaultGetApiKey(), findModel(), discoverAvailableModels()

    · Removed getApiKey callback option (use AuthStorage.setRuntimeApiKey() for runtime overrides)

    · Use getModel() from @mariozechner/pi-ai for built-in models, modelRegistry.find() for custom models + built-in models

    · See updated SDK documentation and README

  • Settings changes: Removed apiKeys from settings.json. Use auth.json instead. (#296)

修复1

  • Duplicate skill warnings for symlinks: Skills loaded via symlinks pointing to the same file are now silently deduplicated instead of showing name collision warnings. (#304 by @mitsuhiko)
0.27.9 2025-12-24 修复 1 Model selector and --list-models with settings.json API keys

修复1

  • Model selector and --list-models with settings.json API keys: Models with API keys configured in settings.json (but not in environment variables) now properly appear in the /model selector and --list-models output. (#295)
0.27.8 2025-12-24 修复 1 API key priority

修复1

  • API key priority: OAuth tokens now take priority over settings.json API keys. Previously, an API key in settings.json would trump OAuth, causing users logged in with a plan (unlimited tokens) to be billed via PAYG instead.
0.27.7 2025-12-24 修复 1 Thinking tag leakage

修复1

  • Thinking tag leakage: Fixed Claude mimicking literal </thinking> tags in responses. Unsigned thinking blocks (from aborted streams) are now converted to plain text without <thinking> tags. The TUI still displays them as thinking blocks. (#302 by @nicobailon)
0.27.6 2025-12-24 新增 2 Compaction hook improvements

新增2

  • Compaction hook improvements: The before_compact session event now includes:

    · previousSummary: Summary from the last compaction (if any), so hooks can preserve accumulated context

    · messagesToKeep: Messages that will be kept after the summary (recent turns), in addition to messagesToSummarize

    · resolveApiKey: Function to resolve API keys for any model (checks settings, OAuth, env vars)

    · Removed apiKey string in favor of resolveApiKey for more flexibility

  • SessionManager API cleanup:

    · Renamed loadSessionFromEntries() to buildSessionContext() (builds LLM context from entries, handling compaction)

    · Renamed loadEntries() to getEntries() (returns defensive copy of all session entries)

    · Added buildSessionContext() method to SessionManager

0.27.5 2025-12-24 新增 2修复 1 HTML export syntax highlighting

新增2

  • HTML export syntax highlighting: Code blocks in markdown and tool outputs (read, write) now have syntax highlighting using highlight.js with theme-aware colors matching the TUI.
  • HTML export improvements: Render markdown server-side using marked (tables, headings, code blocks, etc.), honor user's chosen theme (light/dark), add image rendering for user messages, and style code blocks with TUI-like language markers. (@scutifer)

修复1

  • Ghostty inline images in tmux: Fixed terminal detection for Ghostty when running inside tmux by checking GHOSTTY_RESOURCES_DIR env var. (#299 by @nicobailon)
0.27.4 2025-12-24 修复 1 Symlinked skill directories

修复1

  • Symlinked skill directories: Skills in symlinked directories (e.g., ~/.pi/agent/skills/my-skills -> /path/to/skills) are now correctly discovered and loaded.
0.27.3 2025-12-24 新增 1修复 2 API keys in settings.json

新增1

  • API keys in settings.json: Store API keys in ~/.pi/agent/settings.json under the apiKeys field (e.g., { "apiKeys": { "anthropic": "sk-..." } }). Settings keys take priority over environment variables. (#295)

修复2

  • Allow startup without API keys: Interactive mode no longer throws when no API keys are configured. Users can now start the agent and use /login to authenticate. (#288)
  • --system-prompt file path support: The --system-prompt argument now correctly resolves file paths (like --append-system-prompt already did). (#287 by @scutifer)
0.27.2 2025-12-23 新增 1 Skip conversation restore on branch

新增1

  • Skip conversation restore on branch: Hooks can return { skipConversationRestore: true } from before_branch to create the branched session file without restoring conversation messages. Useful for checkpoint hooks that restore files separately. (#286 by @nicobarray)
0.27.1 2025-12-22 修复 1新增 1 Skill discovery performance

修复1

  • Skill discovery performance: Skip node_modules directories when recursively scanning for skills. Fixes ~60ms startup delay when skill directories contain npm dependencies.

新增1

  • Startup timing instrumentation: Set PI_TIMING=1 to see startup performance breakdown (interactive mode only).
0.27.0 2025-12-22 破坏性变更 1新增 1修复 2 Session hooks API redesign

破坏性变更1

  • Session hooks API redesign: Merged branch event into session event. BranchEvent, BranchEventResult types and pi.on("branch", ...) removed. Use pi.on("session", ...) with reason: "before_branch" | "branch" instead. AgentSession.branch() returns { cancelled } instead of { skipped }. AgentSession.reset() and switchSession() now return boolean (false if cancelled by hook). RPC commands reset, switch_session, and branch now include cancelled in response data. (#278)

新增1

  • Session lifecycle hooks: Added before_* variants (before_switch, before_clear, before_branch) that fire before actions and can be cancelled with { cancel: true }. Added shutdown reason for graceful exit handling. (#278)

修复2

  • File tab completion display: File paths no longer get cut off early. Folders now show trailing / and removed redundant "directory"/"file" labels to maximize horizontal space. (#280)
  • Bash tool visual line truncation: Fixed bash tool output in collapsed mode to use visual line counting (accounting for line wrapping) instead of logical line counting. Now consistent with bash-execution.ts behavior. Extracted shared truncateToVisualLines utility. (#275)
0.26.1 2025-12-22 修复 1 SDK tools respect cwd

修复1

  • SDK tools respect cwd: Core tools (bash, read, edit, write, grep, find, ls) now properly use the cwd option from createAgentSession(). Added tool factory functions (createBashTool, createReadTool, etc.) for SDK users who specify custom cwd with explicit tools. (#279)
0.26.0 2025-12-22 新增 4 SDK for programmatic usage

新增4

  • SDK for programmatic usage: New createAgentSession() factory with full control over model, tools, hooks, skills, session persistence, and settings. Philosophy: "omit to discover, provide to override". Includes 12 examples and comprehensive documentation. (#272)
  • Project-specific settings: Settings now load from both ~/.pi/agent/settings.json (global) and <cwd>/.pi/settings.json (project). Project settings override global with deep merge for nested objects. Project settings are read-only (for version control). (#276)
  • SettingsManager static factories: SettingsManager.create(cwd?, agentDir?) for file-based settings, SettingsManager.inMemory(settings?) for testing. Added applyOverrides() for programmatic overrides.
  • SessionManager static factories: SessionManager.create(), SessionManager.open(), SessionManager.continueRecent(), SessionManager.inMemory(), SessionManager.list() for flexible session management.
0.25.4 2025-12-22 修复 1 Syntax highlighting stderr spam

修复1

  • Syntax highlighting stderr spam: Fixed cli-highlight logging errors to stderr when markdown contains malformed code fences (e.g., missing newlines around closing backticks). Now validates language identifiers before highlighting and falls back silently to plain text. (#274)
0.25.3 2025-12-21 新增 5 Gemini 3 preview models

新增5

  • Gemini 3 preview models: Added gemini-3-pro-preview and gemini-3-flash-preview to the google-gemini-cli provider. (#264 by @LukeFost)
  • External editor support: Press Ctrl+G to edit your message in an external editor. Uses $VISUAL or $EDITOR environment variable. On successful save, the message is replaced; on cancel, the original is kept. (#266 by @aliou)
  • Process suspension: Press Ctrl+Z to suspend pi and return to the shell. Resume with fg as usual. (#267 by @aliou)
  • Configurable skills directories: Added granular control over skill sources with enableCodexUser, enableClaudeUser, enableClaudeProject, enablePiUser, enablePiProject toggles, plus customDirectories and ignoredSkills settings. (#269 by @nicobailon)
  • Skills CLI filtering: Added --skills <patterns> flag for filtering skills with glob patterns. Also added includeSkills setting and glob pattern support for ignoredSkills. (#268)
0.25.2 2025-12-21 修复 1 Image shifting in tool output

修复1

  • Image shifting in tool output: Fixed an issue where images in tool output would shift down (due to accumulating spacers) each time the tool output was expanded or collapsed via Ctrl+O.
0.25.1 2025-12-21 修复 2 Gemini image reading broken

修复2

  • Gemini image reading broken: Fixed the read tool returning images causing flaky/broken responses with Gemini models. Images in tool results are now properly formatted per the Gemini API spec.
  • Tab completion for absolute paths: Fixed tab completion producing //tmp instead of /tmp/. Also fixed symlinks to directories (like /tmp) not getting a trailing slash, which prevented continuing to tab through subdirectories.
0.25.0 2025-12-20 新增 3变更 1修复 3 Interruptible tool execution

新增3

  • Interruptible tool execution: Queuing a message while tools are executing now interrupts the current tool batch. Remaining tools are skipped with an error result, and your queued message is processed immediately. Useful for redirecting the agent mid-task. (#259 by @steipete)
  • Google Gemini CLI OAuth provider: Access Gemini 2.0/2.5 models for free via Google Cloud Code Assist. Login with /login and select "Google Gemini CLI". Uses your Google account with rate limits.
  • Google Antigravity OAuth provider: Access Gemini 3, Claude (sonnet/opus thinking models), and GPT-OSS models for free via Google's Antigravity sandbox. Login with /login and select "Antigravity". Uses your Google account with rate limits.

变更1

  • Model selector respects --models scope: The /model command now only shows models specified via --models flag when that flag is used, instead of showing all available models. This prevents accidentally selecting models from unintended providers. (#255)

修复3

  • Connection errors not retried: Added "connection error" to the list of retryable errors so Anthropic connection drops trigger auto-retry instead of silently failing. (#252)
  • Thinking level not clamped on model switch: Fixed TUI showing xhigh thinking level after switching to a model that doesn't support it. Thinking level is now automatically clamped to model capabilities. (#253)
  • Cross-model thinking handoff: Fixed error when switching between models with different thinking signature formats (e.g., GPT-OSS to Claude thinking models via Antigravity). Thinking blocks without signatures are now converted to text with <thinking> delimiters.
0.24.5 2025-12-20 修复 1 Input buffering in iTerm2

修复1

  • Input buffering in iTerm2: Fixed Ctrl+C, Ctrl+D, and other keys requiring multiple presses in iTerm2. The cell size query response parser was incorrectly holding back keyboard input.
0.24.4 2025-12-20 修复 1 Arrow keys and Enter in selector components

修复1

  • Arrow keys and Enter in selector components: Fixed arrow keys and Enter not working in model selector, session selector, OAuth selector, and other selector components when Caps Lock or Num Lock is enabled. (#243)
0.24.3 2025-12-19 修复 1 Footer overflow on narrow terminals

修复1

  • Footer overflow on narrow terminals: Fixed footer path display exceeding terminal width when resizing to very narrow widths, causing rendering crashes. /arminsayshi
0.24.2 2025-12-20 修复 1 More Kitty keyboard protocol fixes

修复1

  • More Kitty keyboard protocol fixes: Fixed Backspace, Enter, Home, End, and Delete keys not working with Caps Lock enabled. The initial fix in 0.24.1 missed several key handlers that were still using raw byte detection. Now all key handlers use the helper functions that properly mask out lock key bits. (#243)
0.24.1 2025-12-19 新增 2修复 7 OAuth and model config exports

新增2

  • OAuth and model config exports: Scripts using AgentSession directly can now import getAvailableModels, getApiKeyForModel, findModel, login, logout, and getOAuthProviders from @mariozechner/pi-coding-agent to reuse OAuth token storage and model resolution. (#245)
  • xhigh thinking level for gpt-5.2 models: The thinking level selector and shift+tab cycling now show xhigh option for gpt-5.2 and gpt-5.2-codex models (in addition to gpt-5.1-codex-max). (#236 by @theBucky)

修复7

  • Hooks wrap custom tools: Custom tools are now executed through the hook wrapper, so tool_call/tool_result hooks can observe, block, and modify custom tool executions (consistent with hook type docs). (#248 by @nicobailon)
  • Hook onUpdate callback forwarding: The onUpdate callback is now correctly forwarded through the hook wrapper, fixing custom tool progress updates. (#238 by @nicobailon)
  • Terminal cleanup on Ctrl+C in session selector: Fixed terminal not being properly restored when pressing Ctrl+C in the session selector. (#247 by @aliou)
  • OpenRouter models with colons in IDs: Fixed parsing of OpenRouter model IDs that contain colons (e.g., openrouter:meta-llama/llama-4-scout:free). (#242 by @aliou)
  • Global AGENTS.md loaded twice: Fixed global AGENTS.md being loaded twice when present in both ~/.pi/agent/ and the current directory. (#239 by @aliou)
  • Kitty keyboard protocol on Linux: Fixed keyboard input not working in Ghostty on Linux when Num Lock is enabled. The Kitty protocol includes Caps Lock and Num Lock state in modifier values, which broke key detection. Now correctly masks out lock key bits when matching keyboard shortcuts. (#243)
  • Emoji deletion and cursor movement: Backspace, Delete, and arrow keys now correctly handle multi-codepoint characters like emojis. Previously, deleting an emoji would leave partial bytes, corrupting the editor state. (#240)
0.24.0 2025-12-19 新增 7变更 2修复 2破坏性变更 2 Subagent orchestration example

新增7

  • Subagent orchestration example: Added comprehensive custom tool example for spawning and orchestrating sub-agents with isolated context windows. Includes scout/planner/reviewer/worker agents and workflow commands for multi-agent pipelines. (#215 by @nicobailon)
  • getMarkdownTheme() export: Custom tools can now import getMarkdownTheme() from @mariozechner/pi-coding-agent to use the same markdown styling as the main UI.
  • pi.exec() signal and timeout support: Custom tools and hooks can now pass { signal, timeout } options to pi.exec() for cancellation and timeout handling. The result includes a killed flag when the process was terminated.
  • Kitty keyboard protocol support: Shift+Enter, Alt+Enter, Shift+Tab, Ctrl+D, and all Ctrl+key combinations now work in Ghostty, Kitty, WezTerm, and other modern terminals. (#225 by @kim0)
  • Dynamic API key refresh: OAuth tokens (GitHub Copilot, Anthropic OAuth) are now refreshed before each LLM call, preventing failures in long-running agent loops where tokens expire mid-session. (#223 by @kim0)
  • /hotkeys command: Shows all keyboard shortcuts in a formatted table.
  • Markdown table borders: Tables now render with proper top and bottom borders.

变更2

  • Subagent example improvements: Parallel mode now streams updates from all tasks. Chain mode shows all completed steps during streaming. Expanded view uses proper markdown rendering with syntax highlighting. Usage footer shows turn count.
  • Skills standard compliance: Skills now adhere to the Agent Skills standard. Validates name (must match parent directory, lowercase, max 64 chars), description (required, max 1024 chars), and frontmatter fields. Warns on violations but remains lenient. Prompt format changed to XML structure. Removed {baseDir} placeholder in favor of relative paths. (#231)

修复2

  • JSON mode stdout flush: Fixed race condition where pi --mode json could exit before all output was written to stdout, causing consumers to miss final events.
  • Symlinked tools, hooks, and slash commands: Discovery now correctly follows symlinks when scanning for custom tools, hooks, and slash commands. (#219, #232 by @aliou)

破坏性变更2

  • Custom tools now require index.ts entry point: Auto-discovered custom tools must be in a subdirectory with an index.ts file. The old pattern ~/.pi/agent/tools/mytool.ts must become ~/.pi/agent/tools/mytool/index.ts. This allows multi-file tools to import helper modules. Explicit paths via --tool or settings.json still work with any .ts file.
  • Hook tool_result event restructured: The ToolResultEvent now exposes full tool result data instead of just text. (#233)

    · Removed: result: string field

    · Added: content: (TextContent | ImageContent)[] - full content array

    · Added: details: unknown - tool-specific details (typed per tool via discriminated union on toolName)

    · ToolResultEventResult.result renamed to ToolResultEventResult.text (removed), use content instead

    · Hook handlers returning { result: "..." } must change to { content: [{ type: "text", text: "..." }] }

    · Built-in tool details types exported: BashToolDetails, ReadToolDetails, GrepToolDetails, FindToolDetails, LsToolDetails, TruncationResult

    · Type guards exported for narrowing: isBashToolResult, isReadToolResult, isEditToolResult, isWriteToolResult, isGrepToolResult, isFindToolResult, isLsToolResult

0.23.4 2025-12-18 新增 2修复 3 Syntax highlighting

新增2

  • Syntax highlighting: Added syntax highlighting for markdown code blocks, read tool output, and write tool content. Uses cli-highlight with theme-aware color mapping and VS Code-style syntax colors. (#214 by @svkozak)
  • Intra-line diff highlighting: Edit tool now shows word-level changes with inverse highlighting when a single line is modified. Multi-line changes show all removed lines first, then all added lines.

修复3

  • Gemini tool result format: Fixed tool result format for Gemini 3 Flash Preview which strictly requires { output: value } for success and { error: value } for errors. Previous format using { result, isError } was rejected by newer Gemini models. (#213, #220)
  • Google baseUrl configuration: Google provider now respects baseUrl configuration for custom endpoints or API proxies. (#216, #221 by @theBucky)
  • Google provider FinishReason: Added handling for new IMAGE_RECITATION and IMAGE_OTHER finish reasons. Upgraded @google/genai to 1.34.0.
0.23.3 2025-12-17 修复 1变更 3破坏性变更 1 Check for compaction before submitting user prompt, not just

修复1

  • Check for compaction before submitting user prompt, not just after agent turn ends. This catches cases where user aborts mid-response and context is already near the limit.

变更3

  • Improved system prompt documentation section with clearer pointers to specific doc files for custom models, themes, skills, hooks, custom tools, and RPC.
  • Cleaned up documentation:

    · theme.md: Added missing color tokens (thinkingXhigh, bashMode)

    · skills.md: Rewrote with better framing and examples

    · hooks.md: Fixed timeout/error handling docs, added import aliases section

    · custom-tools.md: Added intro with use cases and comparison table

    · rpc.md: Added missing hook_error event documentation

    · README.md: Complete settings table, condensed philosophy section, standardized OAuth docs

  • Hooks loader now supports same import aliases as custom tools (@sinclair/typebox, @mariozechner/pi-ai, @mariozechner/pi-tui, @mariozechner/pi-coding-agent).

破坏性变更1

  • Hooks: turn_end event's toolResults type changed from AppMessage[] to ToolResultMessage[]. If you have hooks that handle turn_end events and explicitly type the results, update your type annotations.
0.23.2 2025-12-17 修复 3 Fixed Claude models via GitHub Copilot re-answering all prev

修复3

  • Fixed Claude models via GitHub Copilot re-answering all previous prompts in multi-turn conversations. The issue was that assistant message content was sent as an array instead of a string, which Copilot's Claude adapter misinterpreted. Also added missing Openai-Intent: conversation-edits header and fixed X-Initiator logic to check for any assistant/tool message in history. (#209)
  • Detect image MIME type via file magic (read tool and @file attachments), not filename extension.
  • Fixed markdown tables overflowing terminal width. Tables now wrap cell contents to fit available width instead of breaking borders mid-row. (#206 by @kim0)
0.23.1 2025-12-17 修复 2 Fixed TUI performance regression caused by Box component lac

修复2

  • Fixed TUI performance regression caused by Box component lacking render caching. Built-in tools now use Text directly (like v0.22.5), and Box has proper caching for custom tool rendering.
  • Fixed custom tools failing to load from ~/.pi/agent/tools/ when pi is installed globally. Module imports (@sinclair/typebox, @mariozechner/pi-tui, @mariozechner/pi-ai) are now resolved via aliases.
0.23.0 2025-12-17 新增 2破坏性变更 1 Custom tools

新增2

  • Custom tools: Extend pi with custom tools written in TypeScript. Tools can provide custom TUI rendering, interact with users via pi.ui (select, confirm, input, notify), and maintain state across sessions via onSession callback. See docs/custom-tools.md and examples/custom-tools/. (#190)
  • Hook and tool examples: Added examples/hooks/ and examples/custom-tools/ with working examples. Examples are now bundled in npm and binary releases.

破坏性变更1

  • Hooks: Replaced session_start and session_switch events with unified session event. Use event.reason ("start" | "switch" | "clear") to distinguish. Event now includes entries array for state reconstruction.
0.22.5 2025-12-17 修复 1 Fixed --session flag not saving sessions in print mode (-p).

修复1

  • Fixed --session flag not saving sessions in print mode (-p). The session manager was never receiving events because no subscriber was attached.
0.22.4 2025-12-17 新增 1修复 1 --list-models [search] CLI flag to list available models wit

新增1

  • --list-models [search] CLI flag to list available models with optional fuzzy search. Shows provider, model ID, context window, max output, thinking support, and image support. Only lists models with configured API keys. (#203)

修复1

  • Fixed tool execution showing green (success) background while still running. Now correctly shows gray (pending) background until the tool completes.
0.22.3 2025-12-16 新增 1变更 2修复 3 Streaming bash output

新增1

  • Streaming bash output: Bash tool now streams output in real-time during execution. The TUI displays live progress with the last 5 lines visible (expandable with ctrl+o). (#44)

变更2

  • Tool output display: When collapsed, tool output now shows the last N lines instead of the first N lines, making streaming output more useful.
  • Updated @mariozechner/pi-ai with X-Initiator header support for GitHub Copilot, ensuring agent calls are not deducted from quota. (#200 by @kim0)

修复3

  • Fixed editor text being cleared during compaction. Text typed while compaction is running is now preserved. (#179)
  • Improved RGB to 256-color mapping for terminals without truecolor support. Now correctly uses grayscale ramp for neutral colors and preserves semantic tints (green for success, red for error, blue for pending) instead of mapping everything to wrong cube colors.
  • /think off now actually disables thinking for all providers. Previously, providers like Gemini with "dynamic thinking" enabled by default would still use thinking even when turned off. (#180 by @markusylisiurunen)
0.22.2 2025-12-15 变更 1 Updated @mariozechner/pi-ai with interleaved thinking enable

变更1

  • Updated @mariozechner/pi-ai with interleaved thinking enabled by default for Anthropic Claude 4 models.
0.22.1 2025-12-15 变更 1 Updated @mariozechner/pi-ai with interleaved thinking suppor

变更1

  • Updated @mariozechner/pi-ai with interleaved thinking support for Anthropic models.
0.22.0 2025-12-15 新增 1修复 1 GitHub Copilot support

新增1

  • GitHub Copilot support: Use GitHub Copilot models via OAuth login (/login -> "GitHub Copilot"). Supports both github.com and GitHub Enterprise. Models are sourced from models.dev and include Claude, GPT, Gemini, Grok, and more. All models are automatically enabled after login. (#191 by @cau1k)

修复1

  • Model selector fuzzy search now matches against provider name (not just model ID) and supports space-separated tokens where all tokens must match
0.21.0 2025-12-14 新增 2修复 2 Inline image rendering

新增2

  • Inline image rendering: Terminals supporting Kitty graphics protocol (Kitty, Ghostty, WezTerm) or iTerm2 inline images now render images inline in tool output. Aspect ratio is preserved by querying terminal cell dimensions on startup. Toggle with /show-images command or terminal.showImages setting. Falls back to text placeholder on unsupported terminals or when disabled. (#177 by @nicobailon)
  • Gemini 3 Pro thinking levels: Thinking level selector now works with Gemini 3 Pro models. Minimal/low map to Google's LOW, medium/high map to Google's HIGH. (#176 by @markusylisiurunen)

修复2

  • Fixed read tool failing on macOS screenshot filenames due to Unicode Narrow No-Break Space (U+202F) in timestamp. Added fallback to try macOS variant paths and consolidated duplicate expandPath functions into shared path-utils.ts. (#181 by @nicobailon)
  • Fixed double blank lines rendering after markdown code blocks (#173 by @markusylisiurunen)
0.20.1 2025-12-13 新增 1 Exported skills API

新增1

  • Exported skills API: loadSkillsFromDir, formatSkillsForPrompt, and related types are now exported for use by other packages (e.g., mom).
0.20.0 2025-12-13 破坏性变更 1新增 1 Pi skills now use SKILL.md convention

破坏性变更1

  • Pi skills now use SKILL.md convention: Pi skills must now be named SKILL.md inside a directory, matching Codex CLI format. Previously any *.md file was treated as a skill. Migrate by renaming ~/.pi/agent/skills/foo.md to ~/.pi/agent/skills/foo/SKILL.md.

新增1

  • Display loaded skills on startup in interactive mode
0.19.1 2025-12-12 修复 1 Documentation: Added skills system documentation to README (

修复1

  • Documentation: Added skills system documentation to README (setup, usage, CLI flags, settings)
0.19.0 2025-12-12 新增 2 Skills system

新增2

  • Skills system: Auto-discover and load instruction files on-demand. Supports Claude Code (~/.claude/skills/*/SKILL.md), Codex CLI (~/.codex/skills/), and Pi-native formats (~/.pi/agent/skills/, .pi/skills/). Skills are listed in system prompt with descriptions, agent loads them via read tool when needed. Supports {baseDir} placeholder. Disable with --no-skills or skills.enabled: false in settings. (#169)
  • Version flag: Added --version / -v flag to display the current version and exit. (#170)
0.18.2 2025-12-11 新增 2修复 3 Auto-retry on transient errors

新增2

  • Auto-retry on transient errors: Automatically retries requests when providers return overloaded, rate limit, or server errors (429, 500, 502, 503, 504). Uses exponential backoff (2s, 4s, 8s). Shows retry status in TUI with option to cancel via Escape. Configurable in settings.json via retry.enabled, retry.maxRetries, retry.baseDelayMs. RPC mode emits auto_retry_start and auto_retry_end events. (#157)
  • HTML export line numbers: Read tool calls in HTML exports now display line number ranges (e.g., file.txt:10-20) when offset/limit parameters are used, matching the TUI display format. Line numbers appear in yellow color for better visibility. (#166)

修复3

  • Branch selector now works with single message: Previously the branch selector would not open when there was only one user message. Now it correctly allows branching from any message, including the first one. This is needed for checkpoint hooks to restore state from before the first message. (#163)
  • In-memory branching for --no-session mode: Branching now works correctly in --no-session mode without creating any session files. The conversation is truncated in memory.
  • Git branch indicator now works in subdirectories: The footer's git branch detection now walks up the directory hierarchy to find the git root, so it works when running pi from a subdirectory of a repository. (#156)
0.18.1 2025-12-10 新增 1修复 1 Mistral provider

新增1

  • Mistral provider: Added support for Mistral AI models. Set MISTRAL_API_KEY environment variable to use.

修复1

  • Fixed print mode (-p) not exiting after output when custom themes are present (theme watcher now properly stops in print mode) (#161)
0.18.0 2025-12-10 新增 6 Hooks system

新增6

  • Hooks system: TypeScript modules that extend agent behavior by subscribing to lifecycle events. Hooks can intercept tool calls, prompt for confirmation, modify results, and inject messages from external sources. Auto-discovered from ~/.pi/agent/hooks/*.ts and .pi/hooks/*.ts. Thanks to @nicobailon for the collaboration on the design and implementation. (#145, supersedes #158)
  • pi.send() API: Hooks can inject messages into the agent session from external sources (file watchers, webhooks, CI systems). If streaming, messages are queued; otherwise a new agent loop starts immediately.
  • --hook <path> CLI flag: Load hook files directly for testing without modifying settings.
  • Hook events: session_start, session_switch, agent_start, agent_end, turn_start, turn_end, tool_call (can block), tool_result (can modify), branch.
  • Hook UI primitives: ctx.ui.select(), ctx.ui.confirm(), ctx.ui.input(), ctx.ui.notify() for interactive prompts from hooks.
  • Hooks documentation: Full API reference at docs/hooks.md, shipped with npm package.
0.17.0 2025-12-09 变更 3新增 2修复 2 Simplified compaction flow

变更3

  • Simplified compaction flow: Removed proactive compaction (aborting mid-turn when threshold approached). Compaction now triggers in two cases only: (1) overflow error from LLM, which compacts and auto-retries, or (2) threshold crossed after a successful turn, which compacts without retry.
  • Compaction retry uses Agent.continue(): Auto-retry after overflow now uses the new continue() API instead of re-sending the user message, preserving exact context state.
  • Merged turn prefix summary: When a turn is split during compaction, the turn prefix summary is now merged into the main history summary instead of being stored separately.

新增2

  • isCompacting property on AgentSession: Check if auto-compaction is currently running.
  • Session compaction indicator: When resuming a compacted session, displays "Session compacted N times" status message.

修复2

  • Block input during compaction: User input is now blocked while auto-compaction is running to prevent race conditions.
  • Skip error messages in usage calculation: Context size estimation now skips both aborted and error messages, as neither have valid usage data.
0.16.0 2025-12-09 破坏性变更 1变更 1 New RPC protocol

破坏性变更1

  • New RPC protocol: The RPC mode (--mode rpc) has been completely redesigned with a new JSON protocol. The old protocol is no longer supported. See C1 for the new protocol documentation and C2 for a working example. Includes RpcClient TypeScript class for easy integration. (#91)

变更1

  • README restructured: Reorganized documentation from 30+ flat sections into 10 logical groups. Converted verbose subsections to scannable tables. Consolidated philosophy sections. Reduced size by ~60% while preserving all information.
0.15.0 2025-12-09 变更 1 Major code refactoring

变更1

  • Major code refactoring: Restructured codebase for better maintainability and separation of concerns. Moved files into organized directories (core/, modes/, utils/, cli/). Extracted AgentSession class as central session management abstraction. Split main.ts and tui-renderer.ts into focused modules. See DEVELOPMENT.md for the new code map. (#153)
0.14.2 2025-12-08 新增 1修复 1 /debug command now includes agent messages as JSONL in the o

新增1

  • /debug command now includes agent messages as JSONL in the output

修复1

  • Fix crash when bash command outputs binary data (e.g., curl downloading a video file)
0.14.1 2025-12-08 修复 1 Fix build errors with tsgo 7.0.0-dev.20251208.1 by properly

修复1

  • Fix build errors with tsgo 7.0.0-dev.20251208.1 by properly importing ReasoningEffort type
0.14.0 2025-12-08 破坏性变更 1新增 4 Custom themes require new color tokens

破坏性变更1

  • Custom themes require new color tokens: Themes must now include thinkingXhigh and bashMode color tokens. The theme loader provides helpful error messages listing missing tokens. See built-in themes (dark.json, light.json) for reference values.

新增4

  • OpenAI compatibility overrides in models.json: Custom models using openai-completions API can now specify a compat object to override provider quirks (supportsStore, supportsDeveloperRole, supportsReasoningEffort, maxTokensField). Useful for LiteLLM, custom proxies, and other non-standard endpoints. (#133, thanks @fink-andreas for the initial idea and PR)
  • xhigh thinking level: Added xhigh thinking level for OpenAI codex-max models. Cycle through thinking levels with Shift+Tab; xhigh appears only when using a codex-max model. (#143)
  • Collapse changelog setting: Add "collapseChangelog": true to ~/.pi/agent/settings.json to show a condensed "Updated to vX.Y.Z" message instead of the full changelog after updates. Use /changelog to view the full changelog. (#148)
  • Bash mode: Execute shell commands directly from the editor by prefixing with ! (e.g., !ls -la). Output streams in real-time, is added to the LLM context, and persists in session history. Supports multiline commands, cancellation (Escape), truncation for large outputs, and preview/expand toggle (Ctrl+O). Also available in RPC mode via {"type":"bash","command":"..."}. (#112, original implementation by @markusylisiurunen)
0.13.2 2025-12-07 变更 1 Tool output truncation

变更1

  • Tool output truncation: All tools now enforce consistent truncation limits with actionable notices for the LLM. (#134)

    · Limits: 2000 lines OR 50KB (whichever hits first), never partial lines

    · read: Shows [Showing lines X-Y of Z. Use offset=N to continue]. If first line exceeds 50KB, suggests bash command

    · bash: Tail truncation with temp file. Shows [Showing lines X-Y of Z. Full output: /tmp/...]

    · grep: Pre-truncates match lines to 500 chars. Shows match limit and line truncation notices

    · find/ls: Shows result/entry limit notices

    · TUI displays truncation warnings in yellow at bottom of tool output (visible even when collapsed)

0.13.1 2025-12-06 新增 1修复 1 Flexible Windows shell configuration

新增1

  • Flexible Windows shell configuration: The bash tool now supports multiple shell sources beyond Git Bash. Resolution order: (1) custom shellPath in settings.json, (2) Git Bash in standard locations, (3) any bash.exe on PATH. This enables Cygwin, MSYS2, and other bash environments. Configure with ~/.pi/agent/settings.json: {"shellPath": "C:\\cygwin64\\bin\\bash.exe"}.

修复1

  • Windows binary detection: Fixed Bun compiled binary detection on Windows by checking for URL-encoded %7EBUN in addition to $bunfs and ~BUN in import.meta.url. This ensures the binary correctly locates supporting files (package.json, themes, etc.) next to the executable.
0.12.15 2025-12-06 修复 1 Editor crash with emojis/CJK characters

修复1

  • Editor crash with emojis/CJK characters: Fixed crash when pasting or typing text containing wide characters (emojis like ✅, CJK characters) that caused line width to exceed terminal width. The editor now uses grapheme-aware text wrapping with proper visible width calculation.
0.12.14 2025-12-06 新增 1 Double-Escape Branch Shortcut

新增1

  • Double-Escape Branch Shortcut: Press Escape twice with an empty editor to quickly open the /branch selector for conversation branching.
0.12.13 2025-12-05 变更 1 Faster startup

变更1

  • Faster startup: Version check now runs in parallel with TUI initialization instead of blocking startup for up to 1 second. Update notifications appear in chat when the check completes.
0.12.12 2025-12-05 变更 1修复 2新增 3 Footer display

变更1

  • Footer display: Token counts now use M suffix for millions (e.g., 10.2M instead of 10184k). Context display shortened from 61.3% of 200k to 61.3%/200k.

修复2

  • Multi-key sequences in inputs: Inputs like model search now handle multi-key sequences identically to the main prompt editor. (#122 by @markusylisiurunen)
  • Line wrapping escape codes: Fixed underline style bleeding into padding when wrapping long URLs. ANSI codes now attach to the correct content, and line-end resets only turn off underline (preserving background colors). (#109)

新增3

  • Fuzzy search models and sessions: Implemented a simple fuzzy search for models and sessions (e.g., codexmax now finds gpt-5.1-codex-max). (#122 by @markusylisiurunen)
  • Prompt History Navigation: Browse previously submitted prompts using Up/Down arrow keys when the editor is empty. Press Up to cycle through older prompts, Down to return to newer ones or clear the editor. Similar to shell history and Claude Code's prompt history feature. History is session-scoped and stores up to 100 entries. (#121 by @nicobailon)
  • /resume Command: Switch to a different session mid-conversation. Opens an interactive selector showing all available sessions. Equivalent to the --resume CLI flag but can be used without restarting the agent. (#117 by @hewliyang)
0.12.11 2025-12-05 变更 1修复 4新增 3 Compaction UI

变更1

  • Compaction UI: Simplified collapsed compaction indicator to show warning-colored text with token count instead of styled banner. Removed redundant success message after compaction. (#108)

修复4

  • Print mode error handling: -p flag now outputs error messages and exits with code 1 when requests fail, instead of silently producing no output.
  • Branch selector crash: Fixed TUI crash when user messages contained Unicode characters (like ✔ or ›) that caused line width to exceed terminal width. Now uses proper truncateToWidth instead of substring.
  • Bash output escape sequences: Fixed incomplete stripping of terminal escape sequences in bash tool output. stripAnsi misses some sequences like standalone String Terminator (ESC \), which could cause rendering issues when displaying captured TUI output.
  • Footer overflow crash: Fixed TUI crash when terminal width is too narrow for the footer stats line. The footer now truncates gracefully instead of overflowing.

新增3

  • authHeader option in models.json: Custom providers can set "authHeader": true to automatically add Authorization: Bearer <apiKey> header. Useful for providers that require explicit auth headers. (#81)
  • --append-system-prompt Flag: Append additional text or file contents to the system prompt. Supports both inline text and file paths. Complements --system-prompt for layering custom instructions without replacing the base system prompt. (#114 by @markusylisiurunen)
  • Thinking Block Toggle: Added Ctrl+T shortcut to toggle visibility of LLM thinking blocks. When toggled off, shows a static "Thinking..." label instead of full content. Useful for reducing visual clutter during long conversations. (#113 by @markusylisiurunen)
0.12.10 2025-12-04 新增 1 Added gpt-5.1-codex-max model support

新增1

  • Added gpt-5.1-codex-max model support
0.12.9 2025-12-04 新增 1 /copy Command

新增1

  • /copy Command: Copy the last agent message to clipboard. Works cross-platform (macOS, Windows, Linux). Useful for extracting text from rendered Markdown output. (#105 by @markusylisiurunen)
0.12.8 2025-12-04 其他 1其他 1 Fix: Use CTRL+O consistently for compaction expand shortcut

未分组1

  • Fix: Use CTRL+O consistently for compaction expand shortcut (not CMD+O on Mac)

未分组1

  • Fix: Use CTRL+O consistently for compaction expand shortcut (not CMD+O on Mac)
0.12.7 2025-12-04 新增 2 Context Compaction

新增2

  • Context Compaction: Long sessions can now be compacted to reduce context usage while preserving recent conversation history. (#92, docs)

    · /compact [instructions]: Manually compact context with optional custom instructions for the summary

    · /autocompact: Toggle automatic compaction when context exceeds threshold

    · Compaction summarizes older messages while keeping recent messages (default 20k tokens) verbatim

    · Auto-compaction triggers when context reaches contextWindow - reserveTokens (default 16k reserve)

    · Compacted sessions show a collapsible summary in the TUI (toggle with o key)

    · HTML exports include compaction summaries as collapsible sections

    · RPC mode supports {"type":"compact"} command and auto-compaction (emits compaction events)

  • Branch Source Tracking: Branched sessions now store branchedFrom in the session header, containing the path to the original session file. Useful for tracing session lineage.
0.12.5 2025-12-03 新增 1修复 1 Forking/Rebranding Support

新增1

  • Forking/Rebranding Support: All branding (app name, config directory, environment variable names) is now configurable via piConfig in package.json. Forks can change piConfig.name and piConfig.configDir to rebrand the CLI without code changes. Affects CLI banner, help text, config paths, and error messages. (#95)

修复1

  • Bun Binary Detection: Fixed Bun compiled binary failing to start after Bun updated its virtual filesystem path format from %7EBUN to $bunfs. (#95)
0.12.4 2025-12-02 新增 1 RPC Termination Safeguard

新增1

  • RPC Termination Safeguard: When running as an RPC worker (stdin pipe detected), the CLI now exits immediately if the parent process terminates unexpectedly. Prevents orphaned RPC workers from persisting indefinitely and consuming system resources.
0.12.3 2025-12-02 修复 2 Rate limit handling

修复2

  • Rate limit handling: Anthropic rate limit errors now trigger automatic retry with exponential backoff (base 10s, max 5 retries). Previously these errors would abort the request immediately.
  • Usage tracking during retries: Retried requests now correctly accumulate token usage from all attempts, not just the final successful one. Fixes artificially low token counts when requests were retried.
0.12.2 2025-12-02 变更 1 Removed support for gpt-4.5-preview and o3 models (not yet a

变更1

  • Removed support for gpt-4.5-preview and o3 models (not yet available)
0.12.1 2025-12-02 新增 1 Models

新增1

  • Models: Added support for OpenAI's new models:

    · gpt-4.1 (128K context)

    · gpt-4.1-mini (128K context)

    · gpt-4.1-nano (128K context)

    · o3 (200K context, reasoning model)

    · o4-mini (200K context, reasoning model)

0.12.0 2025-12-02 新增 8变更 2 -p, --print Flag

新增8

  • -p, --print Flag: Run in non-interactive batch mode. Processes input message or piped stdin without TUI, prints agent response directly to stdout. Ideal for scripting, piping, and CI/CD integration. Exits after first response.
  • -P, --print-streaming Flag: Like -p, but streams response tokens as they arrive. Use --print-streaming --no-markdown for raw unformatted output.
  • --print-turn Flag: Continue processing tool calls and agent turns until the agent naturally finishes or requires user input. Combine with -p for complete multi-turn conversations.
  • --no-markdown Flag: Output raw text without Markdown formatting. Useful when piping output to tools that expect plain text.
  • Streaming Print Mode: Added internal printStreaming option for streaming output in non-TUI mode.
  • RPC Mode print Command: Send {"type":"print","content":"text"} to get formatted print output via print_output events.
  • Auto-Save in Print Mode: Print mode conversations are automatically saved to the session directory, allowing later resumption with --continue.
  • Thinking level options: Added --thinking-off, --thinking-minimal, --thinking-low, --thinking-medium, --thinking-high flags for directly specifying thinking level without the selector UI.

变更2

  • Simplified RPC Protocol: Replaced the prompt wrapper command with direct message objects. Send {"role":"user","content":"text"} instead of {"type":"prompt","message":"text"}. Better aligns with message format throughout the codebase.
  • RPC Message Handling: Agent now processes raw message objects directly, with timestamp auto-populated if missing.
0.11.9 2025-12-02 变更 1 Change Ctrl+I to Ctrl+P for model cycling shortcut to avoid

变更1

  • Change Ctrl+I to Ctrl+P for model cycling shortcut to avoid collision with Tab key in some terminals
0.11.8 2025-12-01 修复 1 Absolute glob patterns (e.g., /Users/foo//.ts) are now handl

修复1

  • Absolute glob patterns (e.g., /Users/foo/**/*.ts) are now handled correctly. Previously the leading / was being stripped, causing the pattern to be interpreted relative to the current directory.
0.11.7 2025-12-01 修复 1 Fix read path traversal vulnerability. Paths are now validat

修复1

  • Fix read path traversal vulnerability. Paths are now validated to prevent reading outside the working directory or its parents. The read tool can read from cwd, its ancestors (for config files), and all descendants. Symlinks are resolved before validation.
0.11.6 2025-12-01 修复 1 Fix --system-prompt <path> allowing the path argument to be

修复1

  • Fix --system-prompt <path> allowing the path argument to be captured by the message collection, causing "file not found" errors.
2025-1113 个版本
0.11.5 2025-11-30 修复 3 Fixed fatal error "Cannot set properties of undefined (setti

修复3

  • Fixed fatal error "Cannot set properties of undefined (setting '0')" when editing empty files in the edit tool.
  • Simplified edit tool output: Shows only "Edited file.txt" for successful edits instead of verbose search/replace details.
  • Fixed fatal error in footer rendering when token counts contain NaN values due to missing usage data.
0.11.4 2025-11-30 修复 1 Fixed chat rendering crash when messages contain preformatte

修复1

  • Fixed chat rendering crash when messages contain preformatted/styled text (e.g., thinking traces with gray italic styling). The markdown renderer now preserves existing ANSI escape codes when they appear before inline elements.
0.11.3 2025-11-29 修复 1 Fix file drop functionality for absolute paths

修复1

  • Fix file drop functionality for absolute paths
0.11.2 2025-11-29 修复 2 Fixed TUI crash when pasting content containing tab characte

修复2

  • Fixed TUI crash when pasting content containing tab characters. Tabs are now converted to 4 spaces before insertion.
  • Fixed terminal corruption after exit when shell integration sequences (OSC 133) appeared in bash output. These sequences are now stripped along with other ANSI codes.
0.11.1 2025-11-29 新增 1修复 1 Added fd integration for file path autocompletion. Now uses

新增1

  • Added fd integration for file path autocompletion. Now uses fd for faster fuzzy file search

修复1

  • Fixed keyboard shortcuts Ctrl+A, Ctrl+E, Ctrl+K, Ctrl+U, Ctrl+W, and word navigation (Option+Arrow) not working in VS Code integrated terminal and some other terminal emulators
0.11.0 2025-11-29 新增 4变更 5修复 3 File-based Slash Commands

新增4

  • File-based Slash Commands: Create custom reusable prompts as .txt files in ~/.pi/slash-commands/. Files become /filename commands with first-line descriptions. Supports {{selection}} placeholder for referencing selected/attached content.
  • /branch Command: Create conversation branches from any previous user message. Opens a selector to pick a message, then creates a new session file starting from that point. Original message text is placed in the editor for modification.
  • Unified Content References: Both @path in messages and --file path CLI arguments now use the same attachment system with consistent MIME type detection.
  • Drag & Drop Files: Drop files onto the terminal to attach them to your message. Supports multiple files and both text and image content.

变更5

  • Model Selector with Search: The /model command now opens a searchable list. Type to filter models by name, use arrows to navigate, Enter to select.
  • Improved File Autocomplete: File path completion after @ now supports fuzzy matching and shows file/directory indicators.
  • Session Selector with Search: The --resume and --session flags now open a searchable session list with fuzzy filtering.
  • Attachment Display: Files added via @path are now shown as "Attached: filename" in the user message, separate from the prompt text.
  • Tab Completion: Tab key now triggers file path autocompletion anywhere in the editor, not just after @ symbol.

修复3

  • Fixed autocomplete z-order issue where dropdown could appear behind chat messages
  • Fixed cursor position when navigating through wrapped lines in the editor
  • Fixed attachment handling for continued sessions to preserve file references
0.10.6 2025-11-28 变更 1修复 3 Show base64-truncated indicator for large images in tool out

变更1

  • Show base64-truncated indicator for large images in tool output

修复3

  • Fixed image dimensions not being read correctly from PNG/JPEG/GIF files
  • Fixed PDF images being incorrectly base64-truncated in display
  • Allow reading files from ancestor directories (needed for monorepo configs)
0.10.5 2025-11-28 新增 1修复 2 Full multimodal support: attach images (PNG, JPEG, GIF, WebP

新增1

  • Full multimodal support: attach images (PNG, JPEG, GIF, WebP) and PDFs to prompts using @path syntax or --file flag

修复2

  • @-references now handle special characters in file names (spaces, quotes, unicode)
  • Fixed cursor positioning issues with multi-byte unicode characters in editor
0.10.4 2025-11-28 修复 1 Removed padding on first user message in TUI to improve visu

修复1

  • Removed padding on first user message in TUI to improve visual consistency.
0.10.3 2025-11-28 新增 1变更 1 Added RPC mode (--rpc) for programmatic integration. Accepts

新增1

  • Added RPC mode (--rpc) for programmatic integration. Accepts JSON commands on stdin, emits JSON events on stdout. See RPC mode documentation for protocol details.

变更1

  • Refactored internal architecture to support multiple frontends (TUI, RPC) with shared agent logic.
0.10.2 2025-11-26 新增 6变更 3 Added thinking level persistence. Default level stored in ~/

新增6

  • Added thinking level persistence. Default level stored in ~/.pi/settings.json, restored on startup. Per-session overrides saved in session files.
  • Added model cycling shortcut: Ctrl+I cycles through available models (or scoped models with -m flag).
  • Added automatic retry with exponential backoff for transient API errors (network issues, 500s, overload).
  • Cumulative token usage now shown in footer (total tokens used across all messages in session).
  • Added --system-prompt flag to override default system prompt with custom text or file contents.
  • Footer now shows estimated total cost in USD based on model pricing.

变更3

  • Replaced --models flag with -m/--model supporting multiple values. Specify models as provider/model@thinking (e.g., anthropic/claude-sonnet-4-20250514@high). Multiple -m flags scope available models for the session.
  • Thinking level border now persists visually after selector closes.
  • Improved tool result display with collapsible output (default collapsed, expand with Ctrl+O).
0.10.1 2025-11-25 新增 1 Add custom model configuration via ~/.pi/models.json

新增1

  • Add custom model configuration via ~/.pi/models.json
0.10.0 2025-11-25 新增 15 Interactive TUI with streaming responses

新增15

  • Interactive TUI with streaming responses
  • Conversation session management with --continue, --resume, and --session flags
  • Multi-line input support (Shift+Enter or Option+Enter for new lines)
  • Tool execution: read, write, edit, bash, glob, grep, think
  • Thinking mode support for Claude with visual indicator and /thinking selector
  • File path autocompletion with @ prefix
  • Slash command autocompletion
  • /export command for HTML session export
  • /model command for runtime model switching
  • /session command for session statistics
  • Model provider support: Anthropic (Claude), OpenAI, Google (Gemini)
  • Git branch display in footer
  • Message queueing during streaming responses
  • OAuth integration for Gmail and Google Calendar access
  • HTML export with syntax highlighting and collapsible sections
这份报告属于 个人工作站 › 调查报告 ← 返回报告清单
返回调查报告