mocode-ai 1.0.13 → 1.1.0

package/README.md

@@ -72,7 +72,7 @@ MoCode isn't a chat box with a coat of paint — it's an agent that actually get
 - **Plan / Auto dual mode** — In `plan` mode the agent is read-only (reads code, queries indexes, searches — never writes to disk, runs commands, or spawns sub-agents) and produces a plan; `auto` mode unlocks the full toolset. The agent can switch between the two on its own — scope out an unfamiliar codebase first, then start making changes.
 - **Automatic context compression** — As the context window fills up, a three-tier compression kicks in (trim individual results → compact older tool results in place → summarize older turns), so long sessions never overflow. `/context` shows live token usage; `/compact` triggers manual compression (optionally with a focus hint to preserve what matters).
 - **Cross-session long-term memory** — The agent can save project architecture, conventions, and lessons learned as long-term memory, auto-loaded in future sessions. A background process periodically reflects on conversations to mine things worth remembering. Memories can be created, searched, updated, and forgotten, with recall-based decay.
-- **Project context (Snapshot + Skill)** — Two complementary systems help the agent understand your project: **Project Snapshot** automatically scans files and generates LLM-enhanced summaries (project description, tech stack, commands, module responsibilities, directory tree); **Project Skill** is a manually maintained knowledge base capturing design decisions, architectural insights, pitfalls, and conventions. Snapshot provides *what/where* (facts), Skill provides *why/how* (insights) — no duplication, ~46% token savings.
+- **Project context (`MOCODE.md`)** — A single project-level memory file at `MOCODE.md` captures both static facts (project description, commands, module list, directory tree) and human/AI-written insights (conventions, architectural decisions, pitfalls). Generate it once with `/init`, then keep it up to date by hand or by asking the agent to refresh it. Loaded automatically into the system prompt on every turn.
 - **Session notepad (notes.md)** — For complex multi-step tasks (≥3 file changes / ≥5 tool calls), the agent maintains a working notepad at `.mocode/sessions/<sessionId>/notes.md` (file-based, survives context compression). It can record intermediate findings, design decisions, open questions, and structured plans. A live progress chip in the TUI status bar shows `plan: [title] (3/7) ▸ [current step]` when a `## Plan:` section is present. The agent manages the file directly with write_file/edit_file/read_file.
 - **Interruptible and reversible** — Ctrl+C interrupts the current turn at any time (kills child processes recursively, rolls history back to before the turn started, leaves no half-finished tool calls). `/rollback` restores file changes from per-turn snapshots, with a per-file keep/undo choice — no git dependency required.
 - **Sandbox protection** — File reads/writes go through a sandbox that blocks out-of-bounds paths (`../../`, absolute paths outside the root, symlink escapes, etc.), so the agent never touches files outside your working directory.
@@ -89,7 +89,7 @@ MoCode isn't a chat box with a coat of paint — it's an agent that actually get
 ## Documentation
 
 - [中文使用指南](./docs/usage.md) — 菜单式快速上手、命令速查、模式、会话、项目上下文与排障。
-- [Project context](./docs/USAGE_SNAPSHOT_SKILL.md) — Snapshot and Project Skill reference.
+- [Project context](./docs/usage.md#项目上下文) — `MOCODE.md` and Skills.
 
 ## Installation
 
@@ -235,8 +235,6 @@ The five `memory_*` tools are gated on `MEMORY_ENABLED=true` at startup; toggle
 | `/pet`              | Toggle the optional desktop pet (floating window mirroring agent state) |
 | `/pet skin`         | Pick a pet skin (↑↓ · Enter)                                          |
 | `/pet quit`         | Fully shut down the pet process (not just disconnect)                 |
-| `/snapshot_refresh` | Refresh project snapshot (re-scan files + regenerate LLM summary)     |
-| `/snapshot`         | Toggle project snapshot on/off                                        |
 
 Type `/` to trigger the dropdown menu, keep typing to filter; Esc to cancel.
 
@@ -264,21 +262,6 @@ MoCode automatically scans the following directories for skills (each skill is a
 
 A skill's `description` is injected into the system prompt (progressive disclosure, tier 1); the model calls `use_skill` to load the full body (tier 2) only when the task is relevant. Use `/skills` to see discovered skills.
 
-## Project Context (Snapshot + Skill)
-
-MoCode uses two complementary systems to help the agent understand your project:
-
-- **Project Snapshot** (automatic + LLM-enhanced, enabled by default): Scans your project files and generates a structured summary including project description, tech stack, key commands, module responsibilities, and directory tree. Stored in `.mocode/snapshot.json` and reused across sessions. Use `/snapshot` to toggle it and `/snapshot_refresh` after major changes.
-- **Project Skill** (manual + AI-assisted, disabled by default): A knowledge base for design decisions, architectural patterns, pitfalls, conventions, and workflow notes. Enable it with `/project_skill on` (or `MOCODE_PROJECT_SKILL=true`), generate an initial draft with `/project_skill init`, then refine `.mocode/project-skill.md`. The agent can update it through `project_skill_update` during conversations.
-
-**Complementary principle**: Snapshot provides *what/where* (facts: files, structure, commands), Skill provides *why/how* (insights: decisions, behaviors, gotchas). No duplication, ~46% token savings compared to the previous approach.
-
-Control via environment variables:
-- `MOCODE_PROJECT_SNAPSHOT=false` — disable snapshot
-- `MOCODE_PROJECT_SKILL=true` — enable Project Skill
-
-See [docs/USAGE_SNAPSHOT_SKILL.md](./docs/USAGE_SNAPSHOT_SKILL.md) for detailed usage.
-
 ## Working discipline (Build-and-Self-Verify)
 
 Every coding task runs through four sequential phases. Skipping or merging them is a failure mode — the system prompt injects this discipline on every turn (see `src/agent/work-discipline.ts`):

package/README.zh-CN.md

@@ -88,7 +88,7 @@ mocode 不是一个套壳聊天框,而是一个能真正动手干活的 agent:
 ## 使用文档
 
 - [菜单式使用指南](./docs/usage.md) — 快速上手、命令速查、模式、会话、项目上下文与排障。
-- [项目上下文：Snapshot 与 Project Skill](./docs/USAGE_SNAPSHOT_SKILL.md)
+- [项目上下文：`MOCODE.md` 与 Skills](./docs/usage.md#项目上下文)
 
 ## 安装
 

package/bin/mocode-agent-host.js

@@ -0,0 +1,3 @@
+#!/usr/bin/env node
+// Mocode Work 的本地 Agent Host。stdout 仅输出 NDJSON 协议事件；诊断写 stderr。
+await import('../dist/host/stdio.js');

package/dist/agent/core.js

@@ -165,7 +165,7 @@ function readDiffContext(tc, parsed) {
  *  - pruner 在每个 runAgentCore 实例化一次(本闭包持有),会话级状态。
  *  - 开关关闭时 pruner 不创建(零开销、零行为变化)。
  *  出口再经 Lifecycle Engine 做引用追踪:LIVE→REFERENCED→OBSOLETE→STUB 四态。
- *  - lifecycle 也在每个 runAgentCore 实例化一次,登记 grep/glob/codegraph 等 producer
+ *  - lifecycle 也在每个 runAgentCore 实例化一次,登记 grep/glob/web_search/web_fetch 等 producer
  *    与 read/edit/write 的 consumer 关系;孤立+老化自动 STUB(观察类工具永不到 STUB)。
  *  - 开关关闭时 lifecycle=null 完全跳过。 */
 /** RETRY-01: 把 thrash hint + 反思指针一次性拼到 output 尾部。
@@ -250,7 +250,8 @@ export async function runAgentCore(opts) {
         ? null
         : (opts.preCompletionChecklist ?? _checklistMiddleware.handler);
     const maxSteps = opts.maxSteps ?? config.maxSteps;
-    // 中断还原:LLM 中途可能调 switch_mode 切了模式,abort 时连同模式一起还原回轮首。
+    // 中断还原:repl 的 /plan / /auto / Shift+Tab 等用户面触发 setAgentMode 中途切了模式,
+    // abort 时连同模式一起还原回轮首。模型不再持有 switch_mode 工具,无法自切。
     const savedMode = getAgentMode();
     // 本轮计时:从入口到完毕(正常 return / 达上限),供 finally 打 ✻ Worked for 摘要行。
     const t0 = Date.now();
@@ -318,7 +319,7 @@ export async function runAgentCore(opts) {
     // 相关性裁剪 pruner:每个 runAgentCore 实例一个,纯静态、不调 LLM、自动判定 read_file 失效。
     // 开关关闭时为 null,所有 pushToolResult 调用走无 pruner 路径(零行为变化)。
     const relprune = config.contextRelprune ? createRelevancePruner() : null;
-    // 观察者生命周期引擎:每个 runAgentCore 实例一个,纯静态、自动维护 grep/glob/codegraph 等
+    // 观察者生命周期引擎:每个 runAgentCore 实例一个,纯静态、自动维护 grep/glob/web_search/web_fetch 等
     // producer 与 read/edit/write 的 consumer 引用关系;孤立+老化的非观察类工具自动 STUB。
     // 开关关闭时为 null,所有 pushToolResult / mutation 调用走无 lifecycle 路径(零行为变化)。
     // 引擎需要从已有会话 history 恢复观察结果的年龄和 path 索引；不能只追踪本次
@@ -693,7 +694,9 @@ export async function runAgentCore(opts) {
                                     : false;
                                 let denied;
                                 if (tool && argumentsValid) {
-                                    const perm = await checkPermission(tool, parsed ?? {}, signal);
+                                    const perm = await checkPermission(tool, parsed ?? {}, signal, {
+                                        prompt: opts.permissionPrompt,
+                                    });
                                     emitTrace('permission', {
                                         source: 'agent_tool',
                                         tool: tc.name,
@@ -785,7 +788,9 @@ export async function runAgentCore(opts) {
                                 ? validateToolArguments(tool, parsed).valid
                                 : false;
                             if (tool && argumentsValid) {
-                                const perm = await checkPermission(tool, parsed ?? {}, signal);
+                                const perm = await checkPermission(tool, parsed ?? {}, signal, {
+                                    prompt: opts.permissionPrompt,
+                                });
                                 emitTrace('permission', {
                                     source: 'agent_tool',
                                     tool: tc.name,

package/dist/agent/mode.js

@@ -2,19 +2,18 @@
  * agent 模式的共享状态(零依赖纯叶子)。
  *
  * 把 `agentMode` 从 repl 的模块变量提到这里,让三方都能用、且不破坏依赖单向:
- *  - `tools/builtins/switch-mode.ts` 写它(setAgentMode)——LLM 在 agent 循环里自切 plan↔auto。
+ *  - `repl/index.ts` 读写它 + 注册 onModeChange 监听器(做 applyMode 重写 history[0] +
+ *    refreshStatusBase 刷状态行 modeTag),并通过 /plan / /auto / Shift+Tab(c cycleMode) 触发切换。
  *  - `agent/index.ts` 每步读它(getAgentMode)——决定 chat() 用全量 chatTools 还是 planChatTools 只读子集,
- *    以及串行分支的 plan 防御 backstop。这样 LLM 中途切模式后,下一次 chat() 立即看到新工具集。
- *  - `repl/index.ts` 读写它 + 注册 onModeChange 监听器(做 applyMode 重写 history[0] + refreshStatusBase
- *    刷状态行 modeTag)。
+ *    以及串行分支的 plan 防御 backstop。这样切换模式后,下一次 chat() 立即看到新工具集。
+ *  - 模型不再持有 switch_mode 工具(已砍):模式切换只能由用户面触发。
  *
- * 依赖方向无环:本模块不 import 任何业务模块。`agent/index → tools/registry → tools/builtins/switch-mode
- * → agent/mode` 链终点是本叶子,不回到 agent/index。`tools → agent/mode` 与既有 `tools → ui`(ask-human)、
- * `tools → rollback`(registry)同性质——都是引叶子。
+ * 依赖方向无环:本模块不 import 任何业务模块。
  *
- * listener 同步触发:setAgentMode 在变更模式后同步调 listener(无 async)。switch_mode 工具在 executeTool
- * 内调 setAgentMode → listener 同步 applyMode(重写 history[0])+ refreshStatusBase;execute 返回字符串后,
- * agent 串行分支继续,下一步 chat() 读到新模式 + 新系统提示,一致。
+ * listener 同步触发:setAgentMode 在变更模式后同步调 listener(无 async)。repl 的
+ * cycleMode / /plan / /auto 在 executeTool 路径之外调 setAgentMode → listener 同步
+ * applyMode(重写 history[0])+ refreshStatusBase;切换后下一步 chat() 读到新模式 +
+ * 新系统提示,一致。
  */
 let currentMode = 'auto';
 let listener = null;

package/dist/agent/spawn.js

@@ -66,7 +66,7 @@ export async function spawnAgent(opts) {
     let toolsOverride;
     const mode = opts.mode ?? 'read';
     const requested = opts.tools?.length ? new Set(opts.tools) : null;
-    const readOnly = new Set(['read_file', 'glob', 'grep', 'codegraph', 'web_search', 'web_fetch', 'use_skill', 'memory_search', 'memory_list']);
+    const readOnly = new Set(['read_file', 'glob', 'grep', 'web_search', 'web_fetch', 'use_skill', 'memory_search', 'memory_list']);
     toolsOverride = chatTools.filter((tool) => tool.function.name !== 'sub-agent' &&
         (!requested || requested.has(tool.function.name)) &&
         (mode === 'write' || readOnly.has(tool.function.name)));
@@ -139,7 +139,7 @@ export async function spawnAgent(opts) {
         onToolOutcome: (tool, args) => {
             if (tool === 'read_file' && typeof args.path === 'string')
                 readSet.add(args.path);
-            else if (['glob', 'grep', 'codegraph'].includes(tool))
+            else if (['glob', 'grep'].includes(tool))
                 readSet.add('workspace');
         },
     });

package/dist/config/index.js

@@ -2,8 +2,6 @@ import fs from 'node:fs';
 import os from 'node:os';
 import path from 'node:path';
 import dotenv from 'dotenv';
-import { loadSnapshot } from '../project-snapshot/index.js';
-import { buildProjectSkillSection } from '../project-skill/index.js';
 import { getSandboxRoot } from '../sandbox/root.js';
 import { getCurrentSessionId } from '../session/state.js';
 import { buildWorkDisciplineSection, inferModelFamily } from '../agent/work-discipline.js';
@@ -85,30 +83,6 @@ const PLATFORM_NOTE = (() => {
  * 默认关(新用户零侵入):这段 + 工具表里的 5 个 memory_* + 系统提示尾部的 Memory Index
  * 都不出现;打开 /memory_switch 后下一次新建 system message 才注入。
  */
-/**
- * 项目快照段落：开 isProjectSnapshotEnabled() 且有快照时才拼。
- * 告诉 LLM 项目已有哪些静态文件可 cache hit，减少无谓 read_file。
- */
-function buildSnapshotSection() {
-    if (!config.projectSnapshotEnabled)
-        return '';
-    try {
-        const snap = loadSnapshot();
-        if (!snap)
-            return '';
-        // 快照内容已经是完整的 markdown，直接返回
-        return `\n${snap.content}\n`;
-    }
-    catch {
-        return '';
-    }
-}
-/**
- * 项目快照总开关查询器（read-file 工具用）。
- */
-export function isProjectSnapshotEnabled() {
-    return config.projectSnapshotEnabled;
-}
 /**
  * Session notepad 段落：读取 .mocode/sessions/<sessionId>/notes.md，只注入 ## 标题行作为目录摘要。
  * Agent 用 write_file/edit_file/read_file 维护此文件，抗 compact（在 context window 之外）。
@@ -165,23 +139,31 @@ const SYSTEM_PROMPT_MEMORY_SECTION = `
  * memoryEnabled=false 时:memory_save/update/forget 三个写工具名字 + "memory-write tools" 这
  * 句都不出现,且 read-only 列表里的 memory_search/memory_list 也移除——避免提示词里出现
  * 根本不存在的工具名引起 LLM 调不到。
+ *
+ * .codegraph/ 存在性也动态决定:有索引时引导 LLM 优先用 codegraph skill(免去逐文件扫);
+ * 没索引时干脆不提,避免 LLM 调出失败。函数化(非 const)以便在 buildPlanModeSuffix 里现拼。
  */
-const PLAN_RESEARCH_RULES = `
-- Research enough to locate relevant code, trace call paths, and understand existing conventions. Use the codegraph-first Workflow, but do not repeat information already retrieved in this session.
+function buildPlanResearchRules() {
+    const cg = hasCodegraphIndex()
+        ? ' If `.codegraph/` exists, prefer loading the `codegraph` skill (via use_skill) and using run_command to query it.'
+        : '';
+    return `
+- Research enough to locate relevant code, trace call paths, and understand existing conventions.${cg} Otherwise rely on read_file / glob / grep. Do not repeat information already retrieved in this session.
 - Produce an actionable plan: files and reasons, ordered steps, edge cases, and verification (typecheck / tests / build).
 - When ready, MUST call the \`ask_human\` tool with a concise summary and exactly these options:
-  1. "按计划执行 (switch to auto and implement)" — call \`switch_mode("auto")\` in the same turn, then implement.
+  1. "按计划执行 (please run /auto to switch to auto mode and proceed)" — wait for the user to run /auto, then implement.
   2. "继续细化方案 (stay in plan, refine)" — remain in plan and refine.
   3. "取消 / 暂不执行 (abort)" — stop without switching mode.
 - Never silently switch or stop. Do not ask approval in plain text; \`ask_human\` is the approval channel.
 - The REPL approval prompt is only a fallback; do not rely on it.`;
+}
 function buildPlanModeSuffix() {
     const memoryTools = isMemoryEnabled()
         ? 'memory_save, memory_update, memory_forget'
         : '';
     const readOnlyTools = isMemoryEnabled()
-        ? 'read_file, glob, grep, codegraph, web_search, web_fetch, use_skill, ask_human, memory_search, memory_list'
-        : 'read_file, glob, grep, codegraph, web_search, web_fetch, use_skill, ask_human';
+        ? 'read_file, glob, grep, web_search, web_fetch, use_skill, ask_human, memory_search, memory_list'
+        : 'read_file, glob, grep, web_search, web_fetch, use_skill, ask_human';
     const removed = ['write_file', 'edit_file', 'run_command', memoryTools]
         .filter(Boolean)
         .join(', ');
@@ -190,7 +172,7 @@ function buildPlanModeSuffix() {
 ## ⛯ PLAN MODE (active now)
 You are in PLAN mode: investigate and design only — do NOT execute or change anything.
 - Removed from your tool list: ${removed}. Use only these read-only tools: ${readOnlyTools}.
-${PLAN_RESEARCH_RULES}`;
+${buildPlanResearchRules()}`;
 }
 /** 兼容旧名字:repl 的 buildSystemMessage 仍引 PLAN_MODE_SUFFIX(变量)。运行时按需现拼。 */
 export function buildBasePrompt(sessionId = getCurrentSessionId()) {
@@ -225,12 +207,12 @@ ${buildWorkDisciplineSection(inferModelFamily(config.model))}
 
 ## Workflow
 - Understand requirements and current code before acting; do not guess.
-- If \`.codegraph/\` exists, use \`codegraph\` first for unfamiliar code questions. Use direct reads for known or recently changed files.
 - After modifications, run the smallest relevant verification, then typecheck/build when appropriate. Never claim success without evidence.
 - Use web search only when freshness materially affects the answer (new APIs, versions, security, current UI conventions).
+${buildCodegraphSection()}
 
 ## Tool rules
-- Precise path/symbol → go directly to \`read_file\` or \`codegraph node\`; use \`glob\`/\`grep\` only for discovery.
+- Precise path/symbol → go directly to \`read_file\`; use \`glob\`/\`grep\` only for discovery. If a codegraph index is available (see Workflow above), prefer \`codegraph node\` / \`codegraph query\` for symbol lookups.
 - Before editing, read the exact target region and copy both its artifact \`hash\` and verbatim text. Use \`edit_file\` with \`expected_hash\` for unique local replacements, and \`write_file\` with the latest hash for replacement (or null only for creation).
 - Local edits require an exact unique match; use \`write_file\` for new/full files.
 - Use \`glob\`/\`grep\` for discovery and \`run_command\` for execution or verification, not file existence checks. State intent before side effects.
@@ -254,7 +236,7 @@ ${buildWorkDisciplineSection(inferModelFamily(config.model))}
 - Operate only within authorized scope; when unsure, ask — don't guess.
 
 ## Project context (dynamic reference)
-${buildSnapshotSection()}${config.projectSkillEnabled ? buildProjectSkillSection() : ''}${memorySection}${buildNotepadSection(sessionId)}
+${memorySection}${buildNotepadSection(sessionId)}
 
 ## Session Notepad — working notes file
 ${sessionId
@@ -394,10 +376,8 @@ export const config = {
     theme: process.env.MOCODE_THEME || 'default',
     themeFromShell,
     llmKeysFromShell,
-    projectSnapshotEnabled: process.env.MOCODE_PROJECT_SNAPSHOT !== 'false',
     permissionEnabled: process.env.MOCODE_PERMISSION !== 'false',
     permissionNonInteractiveAllow: process.env.MOCODE_PERMISSION_NON_INTERACTIVE_ALLOW === 'true',
-    projectSkillEnabled: process.env.MOCODE_PROJECT_SKILL === 'true',
 };
 /**
  * 运行时更新模型相关配置(/model 命令调)。
@@ -442,6 +422,32 @@ export function updateSubAgentConfig(enabled) {
 export function isMemoryEnabled() {
     return config.memoryEnabled;
 }
+/**
+ * .codegraph/ 索引存在性:仅查 cwd 顶层 .codegraph(目录或文件均可,codegraph CLI
+ * 自己会处理内部布局)。用于动态决定是否在系统提示里注入 codegraph skill 用法段——
+ * 没有索引的项目不应被提示「用 codegraph」以免 LLM 调出失败。失败静默返 false。
+ */
+export function hasCodegraphIndex() {
+    try {
+        return fs.existsSync(path.join(process.cwd(), '.codegraph'));
+    }
+    catch {
+        return false;
+    }
+}
+/**
+ * 当 .codegraph/ 存在时拼进 auto 模式系统提示的 codegraph 段;否则返空串(零成本)。
+ * 单一来源:被 buildBasePrompt 注入,确保 basePrompt 不含死字符串。
+ */
+export function buildCodegraphSection() {
+    if (!hasCodegraphIndex())
+        return '';
+    return [
+        '',
+        '## Codegraph (project has .codegraph/ index)',
+        '- For unfamiliar code questions, prefer loading the `codegraph` skill (via use_skill) and querying it with run_command (`codegraph explore <entry>`, `codegraph node <symbol>`). Falls back to read_file / glob / grep when not applicable.',
+    ].join('\n');
+}
 /**
  * 切换记忆子系统开关(/memory_switch on|off 调)。
  * - 更新 config 单例字段(其它模块下次调 isMemoryEnabled() 即拿新值)。
@@ -457,34 +463,6 @@ export function updateMemoryConfig(enabled) {
     config.memoryEnabled = enabled;
     process.env.MEMORY_ENABLED = enabled ? 'true' : 'false';
 }
-/**
- * 项目专属 Skill 总开关:单一来源。/project_skill、buildBasePrompt、
- * tools/builtins/index.ts 都从这里查。默认 false(零侵入)。
- */
-export function isProjectSkillEnabled() {
-    return config.projectSkillEnabled;
-}
-/**
- * 切换项目专属 Skill 开关(/project_skill on|off 调)。
- * - 更新 config 单例字段(其它模块下次调 isProjectSkillEnabled() 即拿新值)。
- * - 同步 process.env.MOCODE_PROJECT_SKILL(下次启动 loadEnvFiles 不被文件回填)。
- * 持久化(写 ~/.mocode/config 的 MOCODE_PROJECT_SKILL 键)由调用方走 writeConfigKeys。
- */
-export function updateProjectSkillConfig(enabled) {
-    config.projectSkillEnabled = enabled;
-    process.env.MOCODE_PROJECT_SKILL = enabled ? 'true' : 'false';
-}
-/**
- * 切换项目快照开关(/snapshot on|off 调)。
- * - 更新 config 单例字段(其它模块下次读 config.projectSnapshotEnabled 即拿新值:
- *   buildSnapshotSection 现拼现读、read-file 每次 execute 现读)。
- * - 同步 process.env.MOCODE_PROJECT_SNAPSHOT(下次启动 loadEnvFiles 不会被文件回填)。
- * 持久化(写 ~/.mocode/config 的 MOCODE_PROJECT_SNAPSHOT 键)由调用方走 updateConfigKey。
- */
-export function updateSnapshotConfig(enabled) {
-    config.projectSnapshotEnabled = enabled;
-    process.env.MOCODE_PROJECT_SNAPSHOT = enabled ? 'true' : 'false';
-}
 /** 切换界面与模型回复语言；持久化由 REPL 调用 config/file.ts 完成。 */
 export function updateLanguageConfig(language) {
     setLanguage(language);

package/dist/context/artifacts.js

@@ -40,7 +40,7 @@ function callArgs(history, idx) {
 function sourceType(tool) {
     if (tool === 'read_file')
         return 'read';
-    if (tool === 'grep' || tool === 'glob' || tool === 'codegraph')
+    if (tool === 'grep' || tool === 'glob')
         return 'search';
     if (tool === 'run_command')
         return 'diagnostic';
@@ -53,7 +53,7 @@ function pathsFromOutput(tool, output) {
         if (normalized)
             paths.add(normalized);
     };
-    if (tool === 'grep' || tool === 'codegraph' || tool === 'run_command') {
+    if (tool === 'grep' || tool === 'run_command') {
         const expression = /^(.+?\.[A-Za-z0-9]+):(?:\d+|\s*\d+\s*(?:处匹配|matches?))/gmi;
         let match;
         while ((match = expression.exec(output)))

package/dist/context/classifier.js

@@ -1,21 +1,19 @@
 // Context Classifier:据工具名(强先验)+ 输出形状(启发)+ 兜底,选 ContextKind。
 //
 // 三级信号:
-//  1) 名字强先验(BY_NAME 表,覆盖全部 17 内置工具,确定性强)。
+//  1) 名字强先验(BY_NAME 表,覆盖全部内置工具,确定性强)。
 //  2) 形状启发(为 MCP 工具 / 未来工具 / 未登记工具兜底识别)。
 //  3) 兜底 'passthrough'(不认识 = 不动,零行为变化)。
 //
 // 单一事实源风格(仿 tools/constants.ts 的 READ_TOOL_NAMES / PLAN_DISABLED_TOOLS)。
 // 加新工具:在 BY_NAME 加一行;或靠形状启发自动识别。
-/** 工具名 → ContextKind 的强先验表(覆盖全部 18 内置工具)。 */
+/** 工具名 → ContextKind 的强先验表(覆盖全部内置工具)。 */
 const BY_NAME = {
     // tree:路径列表 → 缩进树
     glob: 'tree',
     // search:file:line 分组
     grep: 'search',
     web_search: 'search',
-    // graph:CLI dump → 精炼图
-    codegraph: 'graph',
     // log:分级 / 折叠 / 尾偏置
     run_command: 'log',
     // code:保行号(edit_file 依赖,最敏感)
@@ -31,7 +29,6 @@ const BY_NAME = {
     edit_file: 'status',
     write_file: 'status',
     ask_human: 'status',
-    switch_mode: 'status',
     drop_context: 'status',
     memory_save: 'status',
     memory_update: 'status',
@@ -47,7 +44,7 @@ function classifyByShape(output) {
     // file:line: content 形(grep 风格)
     if (/^[^\n:]+:\d+:[^\n]*$/m.test(output))
         return 'search';
-    // [退出码 N] 前缀(run_command / codegraph 风格)
+    // [退出码 N] 前缀(run_command 风格)
     if (/^\[退出码 \d+\]/m.test(output))
         return 'log';
     // 路径列表:多行都是含分隔符的相对路径(glob 风格)

package/dist/context/encoders/_util.js

@@ -6,8 +6,9 @@ export function stripAnsi(s) {
 }
 /**
  * 折叠连续空行(只含空白字符的行)≥ threshold → 单个空行。
- * 用于 prose / source dump(web_fetch / use_skill / task / codegraph):HTML→文本与 markdown 常留多余空行,
+ * 用于 prose / source dump(web_fetch / use_skill / task):HTML→文本与 markdown 常留多余空行,
  * 多空行与单空行语义等价,折叠无损。默认 threshold=3(只动真正过量的空行,常见 ≤2 空行不动)。
+ * 注:codegraph 已 skill 化(用 run_command 调 CLI,落到 log encoder),graph kind 现仅作保留值。
  *
  * 注意:用 `.trim() === ''` 判空——故 read_file 的 `     2\t`(行号前缀 + tab,trim 后剩 `2`)不会被
  * 视作空行。read_file 的空行折叠见 code encoder(前缀感知)。

package/dist/context/encoders/index.js

@@ -4,8 +4,9 @@
 // Phase 2:tree / search / log / table / memory(高价值低风险)。
 // Phase 2.5:code / graph / doc / summary(覆盖剩余有 encoder 的 kind)。
 //   - code(read_file):仅折叠 ≥3 连续空行,行号保真(edit_file 依赖)。
-//   - graph(codegraph)/ doc(web_fetch, use_skill)/ summary(task):去 ANSI + 折叠空行,保守不重构结构。
-//   - status(edit/write/ask_human/switch_mode/mem 增删改)无 encoder:本就是一行,无需编码。
+//   - graph(保留 kind,暂无 builtin 工具直接命中)/ doc(web_fetch, use_skill)/ summary(task):
+//     去 ANSI + 折叠空行,保守不重构结构。
+//   - status(edit/write/ask_human/mem 增删改)无 encoder:本就是一行,无需编码。
 //
 // 加 encoder:新建 encoders/xxx.ts 导出 ContextEncoder,在此数组加一行。无需动 agent / llm / core。
 import { passthroughEncoder } from './passthrough.js';
@@ -15,7 +16,6 @@ import { commandEncoder } from './command.js';
 import { tableEncoder } from './table.js';
 import { memoryEncoder } from './memory.js';
 import { codeEncoder } from './code.js';
-import { graphEncoder } from './graph.js';
 import { docEncoder } from './doc.js';
 import { summaryEncoder } from './summary.js';
 export const builtinEncoders = [
@@ -26,7 +26,6 @@ export const builtinEncoders = [
     tableEncoder,
     memoryEncoder,
     codeEncoder,
-    graphEncoder,
     docEncoder,
     summaryEncoder,
 ];

package/dist/context/lifecycle.js

@@ -1,7 +1,7 @@
 // Observation Lifecycle Engine:tool 消息的「观察者生命周期」状态机。
 //
 // 在 Relevance Pruner 之上的第二层被动裁剪。Relevance Pruner 只管 read_file 的「同 path 新旧替换 +
-// mutation 覆写」,本层补足「grep/glob/codegraph 这类观察类工具」的引用追踪。
+// mutation 覆写」,本层补足「grep/glob/web_search/web_fetch 这类观察类工具」的引用追踪。
 //
 // 四态机(LIVE → REFERENCED → OBSOLETE → STUB):
 //   - LIVE:刚 push 进 history 的工具结果,尚未被任何下游工具消费。
@@ -10,7 +10,7 @@
 //   - STUB:已被替换为存根(物理上 content 变成「⌦[无消费者:...]」)。
 //
 // 观察类工具两阶段衰减(避免误伤):
-//   - grep/glob/codegraph/web_search/web_fetch 等「观察/检索类」工具两阶段衰减:
+//   - grep/glob/web_search/web_fetch 等「观察/检索类」工具两阶段衰减:
 //     Phase 1(10 步):LIVE → REFERENCED,保留完整内容(返回多个候选,剩余候选可能后续被消费)。
 //     Phase 2(+5 步):REFERENCED → DIGEST,替换为摘要存根(保留文件列表+命中数+参数,丢弃详情),
 //     释放 ~90% token。states 仍为 REFERENCED,不引入新状态。
@@ -34,7 +34,6 @@ import { canonicalizePath, extractPath, isToolResultSuccess, lastUserIndex, toTe
 const OBSERVER_TOOLS = new Set([
     'grep',
     'glob',
-    'codegraph',
     'web_search',
     'web_fetch',
 ]);
@@ -63,7 +62,6 @@ const OBSERVER_DIGEST_AGE = 5;
  *  - grep:content 是 `file:line: ...` 行,提取每行的 file 段(只保留绝对路径形态或与 pattern 匹配的)。
  *    简化:把所有看起来像「相对路径 + 文件名」的 token 抽出,留 narrow。
  *  - glob:content 是路径列表,按行 / 空格拆。
- *  - codegraph:content 里通常含 `path/to/file.ts:line`,按行拆,提 file 段。
  *
  *  返回值:命中过的 path 字符串集合(已 dedup)。失败返空集。 */
 function extractProducerPaths(toolName, content) {
@@ -99,14 +97,6 @@ function extractProducerPaths(toolName, content) {
                     addPath(t);
             }
         }
-        else if (toolName === 'codegraph') {
-            // codegraph 输出通常 `path\to\file.ts:line:col  symbol` 或类似;按行 + 冒号分隔。
-            for (const line of content.split(/\r?\n/)) {
-                const m = /^([^\s:][^:]*?\.[A-Za-z0-9]+):(\d+):/.exec(line);
-                if (m)
-                    addPath(m[1]);
-            }
-        }
         else if (toolName === 'web_search' || toolName === 'web_fetch') {
             // 网络结果不在文件系统路径范畴;不参与 producer 路径索引(避免误匹配)。
         }
@@ -287,7 +277,7 @@ export class LifecycleEngine {
                 })();
                 const path = canonicalizePath(extractPath(argsRaw));
                 if (path) {
-                    // 1) 找该 path 的所有上游 producer(grep/glob/codegraph)→ 标 REFERENCED。
+                    // 1) 找该 path 的所有上游 producer(grep/glob/web_search/web_fetch)→ 标 REFERENCED。
                     const producers = this.producersByPath.get(path);
                     if (producers) {
                         for (const pidx of producers) {
@@ -478,16 +468,6 @@ export class LifecycleEngine {
                 } })();
                 return `${DIGEST_PREFIX}glob(${args}) — 历史结果文件 ${files}，${origLen}→摘要]\n如需完整列表或最新信息请重新调用 glob`;
             }
-            case 'codegraph': {
-                const args = (() => { try {
-                    const a = JSON.parse(argsRaw);
-                    return `"${a.query ?? ''}"`;
-                }
-                catch {
-                    return '...';
-                } })();
-                return `${DIGEST_PREFIX}codegraph(${args}) — 历史结果文件 ${files}，${origLen}→摘要]\n如需完整详情或最新信息请重新调用 codegraph`;
-            }
             case 'web_search': {
                 // 统计结果数
                 let resultCount = 0;

package/dist/context/relevance.js

@@ -35,21 +35,6 @@ function observationKey(call) {
             maxPerFile: Math.min(Math.max(rawMax, 1), 50),
         });
     }
-    if (call.name === 'codegraph') {
-        if (typeof args.action !== 'string' || typeof args.query !== 'string')
-            return null;
-        const query = args.action === 'explore'
-            ? args.query.trim().replace(/\s+/g, ' ')
-            : args.query.trim();
-        return JSON.stringify({
-            tool: 'codegraph',
-            action: args.action,
-            query,
-            file: typeof args.file === 'string' ? args.file.trim().replace(/\\/g, '/') : '',
-            offset: args.offset === undefined ? null : normalizedInteger(args.offset, 0),
-            limit: args.limit === undefined ? null : normalizedInteger(args.limit, 0),
-        });
-    }
     return null;
 }
 function observationLabel(call) {
@@ -59,14 +44,12 @@ function observationLabel(call) {
         const glob = JSON.stringify(String(args.glob ?? '**/*')).slice(0, 80);
         return `grep(pattern=${pattern}, glob=${glob})`;
     }
-    const action = String(args.action ?? '');
-    const query = JSON.stringify(String(args.query ?? '')).slice(0, 100);
-    return `codegraph(action=${action}, query=${query})`;
+    return call.name;
 }
 /**
  * Cross-message relevance pruning:
  * - read_file: a newer successful read of the same canonical path supersedes old reads.
- * - grep/codegraph: a newer successful call with the exact same semantic arguments
+ * - grep: a newer successful call with the exact same semantic arguments
  *   supersedes old results. Partial file overlap is intentionally not enough.
  */
 export class RelevancePruner {
@@ -186,9 +169,7 @@ export class RelevancePruner {
                 return;
             if (!message.tool_call_id)
                 return;
-            const reason = toolName === 'grep'
-                ? '相同 grep 查询已有更新结果'
-                : '相同 codegraph 查询已有更新结果';
+            const reason = '相同 grep 查询已有更新结果';
             message.content =
                 `${STUB_PREFIX}${reason}] ${observationLabel(call)} ${content.length} 字符 ` +
                     `→ 已被更新查询替代 · id …${message.tool_call_id.slice(-6)}⌫`;

package/dist/host/protocol.js

@@ -0,0 +1,40 @@
+function parseAttachments(value) {
+    if (!Array.isArray(value))
+        return undefined;
+    return value.flatMap((item) => {
+        if (!item || typeof item !== 'object')
+            return [];
+        const entry = item;
+        if (typeof entry.name !== 'string' || typeof entry.dataUrl !== 'string' || !entry.dataUrl.startsWith('data:image/'))
+            return [];
+        return [{ name: entry.name.slice(0, 240), dataUrl: entry.dataUrl }];
+    });
+}
+export function parseCommand(value) {
+    if (!value || typeof value !== 'object')
+        return null;
+    const input = value;
+    if (typeof input.id !== 'string' || typeof input.type !== 'string')
+        return null;
+    if (input.type === 'run' && typeof input.prompt === 'string') {
+        return {
+            id: input.id,
+            type: 'run',
+            prompt: input.prompt,
+            sessionId: typeof input.sessionId === 'string' ? input.sessionId : undefined,
+            attachments: parseAttachments(input.attachments),
+        };
+    }
+    if (input.type === 'cancel')
+        return { id: input.id, type: 'cancel' };
+    if (input.type === 'approval' && typeof input.approvalId === 'string') {
+        return {
+            id: input.id,
+            type: 'approval',
+            approvalId: input.approvalId,
+            action: input.action === 'selected' ? 'selected' : 'cancelled',
+            value: typeof input.value === 'string' ? input.value : undefined,
+        };
+    }
+    return null;
+}