thincoder 0.12.61 → 0.12.63

package/CHANGELOG.md

@@ -1,3 +1,36 @@
+## [0.12.63] — 2026-09-21
+
+> 0.12.62 → 0.12.63（连续号——发布时定号；本版随核首发联动）
+
+### Fixed
+
+- **发布依赖收正**：`@thincoder/core` `^0.1.0` → `^0.9.1`（核首发 `@thincoder/core@0.9.1` 同日落地——声明面与 registry 实发对齐；0.12.62 及以前发布形态为合并前自带拷贝、无核依赖）。
+- **smoke 探针 import 修复**：`isBailianHost` 实居 `config.mjs`——原从 `provider/normalize.mjs` 静默容错导入（`catch(() => ({}))`）恒 undefined ⇒ smoke 恒 skip（假绿）✗ 修正后真跑三连测（OFF 无思考 / XHIGH / DEFAULT 有思考——dashscope 百炼 qwen 实测过 ✓）。
+
+## [0.12.62] — 2026-09-13
+
+> 0.12.61 → 0.12.62（§4.3 连续号——发布时定号）
+
+### Added
+
+- **文档锚一致性机检 V5（DOC-CODE-RECONCILE 批）**：`scripts/doc-anchors.mjs`（三类锚——用例号 / 本仓符号 / 路径——存在性判定 + 假阳类逐条排除 + 注记闭枚举 + 两态 `V5_GATE` / `--v5-gate` / `--v5-report`）· `scripts/doc-impact.mjs`（改动面反查：`--base` 必给、只读不阻断）· 常驻零值锁（`test:full` 内——再出现即红）· **清账 347 处 → 0**。
+- **台账自持机检**：`check-ledger` L4 本仓可解析闸 + 指针 / D3 组计数 / 六态 / 触发 / 老化报告。
+- **工程模式纪律进提示词**：跨仓派单与写域纪律 · 改动面反查（提示词两副本逐字一致）。
+
+### Changed
+
+- **文档体系各仓自持**：需求档 / 设计档 / 批次档 / 台账各仓记各仓（禁跨仓写需求与跨仓指针）；跨仓引用统一「名称（仓别）§N」规范形态。
+- **台账归档制**：活文件只留未决四态，已决逐条移入 `TODO-archive.md`。
+- **在途批一并签入（8 批）**：TUI 内存取证与上界（session store / segments / 堆遥测）· 台账可见面 · 测试纪律提示词 · 子代理审批面与活动区对位。
+
+### Removed
+
+- **散文锚退役（PROSE-ANCHOR-RETIRE 批）**：退役断言「某句在场 / 缺席」类测试 71 例（保留结构机检面：引用可解析 / 组计数 / 六态取值 / 行宽 / slow 归册）。
+
+### Fixed
+
+- 文档↔实装漂移清账（逐档处置留痕，禁「只删不记」）· `check-ledger` 扫描域收窄（只扫本仓活档 + 归档档）· 台账 L4 误报修正 · 批次档 / 设计档宽度与计数纪律收正。
+
 ## [0.12.61] — 2026-09-11
 
 > 0.12.60 → 0.12.61（§4.3 连续号——发布时定号）
@@ -291,7 +324,7 @@
 
 ### Changed
 
-- **系统提示词加两条约束**：①「先定正确再谈实现」——动手前先回答"正确应该是什么"（每个入口/视图/边界），再谈怎么实现，实现规模是"正确"的结果、不是决策标准；②「确认时暴露方案权衡」——确认理解时列出选的做法、为什么它对（绝不是"改动最小"）、考虑过但没选的替代方案。治"最小改动"偷懒病根
+- **系统提示词加两条约束**：①「先定正确再谈实现」——动手前先回答"正确应该是什么"（每个入口/视图/边界），再谈怎么实现，实现规模是"正确"的结果、不是决策标准；②「确认时暴露方案权衡」——确认理解时列出选的做法、为什么它对（判据 = 正确性优先——不是省事程度）、考虑过但没选的替代方案。治"图省事"的偷懒病根
 
 ## [0.12.47] — 2026-08-28
 

package/README.md

@@ -15,12 +15,12 @@ Design philosophy (the entire meaning of the name): if the Node standard library
 - **Fix-verify loop**: file changes without `verify` get pushed back — syntax check + tests must pass before the agent can claim completion (auto-repair up to 3 rounds)
 - **Checkpoint system**: auto-snapshot before every user task, `list`/`create`/`rewind` tools for the model, single-file restore — rewinding itself is reversible (pre-rewind state auto-saved)
 - **Codebase understanding** ⭐0.5.0: `repo_outline` (dependency outline, auto-injected at startup), `code_search` (source FTS5 + vectors + JSDoc extraction), `doc_search` (docs chunked by ## headings) — background indexing, auto-incremental updates on file writes, three tools guided by "structure → intent → details"
-- **Model adaptation** ⭐: top-tier only, latest only. Built-in flagship models from twenty providers — DeepSeek / Kimi / Kimi For Coding / GLM / GLM Coding Plan (智谱 Coding) / Qwen / Qwen Token Plan / MiMo (小米) / MiMo Token Plan (小米套餐) / MiniMax / OpenAI / Claude / Gemini / Grok / Mistral / Volcengine Ark (豆包) / Hunyuan (腾讯混元) / SiliconFlow (硅基流动) / OpenRouter / Groq. No legacy model compatibility. Auto-matched context windows, truncation-resume protocols (prefix/partial), thinking-mode APIs (thinking.type / reasoning_effort), reasoning_content echo strategies (reasoningEcho), output limits, temperature range clamping — all deeply adapted.
+- **Model adaptation** ⭐: top-tier only, latest only. Built-in flagship models from twenty-one providers — DeepSeek / Kimi / Kimi For Coding / GLM / GLM Coding Plan (智谱 Coding) / Qwen / Qwen Token Plan / MiMo (小米) / MiMo Token Plan (小米套餐) / MiniMax / OpenAI / Claude / Gemini / Grok / Mistral / Volcengine Ark (豆包) / Hunyuan (腾讯混元) / Tencent TokenHub (腾讯混元网关) / SiliconFlow (硅基流动) / OpenRouter / Groq. No legacy model compatibility. Auto-matched context windows, truncation-resume protocols (prefix/partial), thinking-mode APIs (thinking.type / reasoning_effort), reasoning_content echo strategies (reasoningEcho), output limits, temperature range clamping — all deeply adapted.
 - **Toolset**: `read` / `write` / `edit` / `bash` / `glob` (supports `**`) / `grep` / `websearch` / `ls` / `fetch` + `read_image` (image/video paste) + three retrieval tools + MCP — all zero-dependency, file tools confined to the working directory
 - **Memory system**: three layers (personal/project/team), FTS5 + vector RRF hybrid retrieval, git-friendly markdown format
 - **Two-phase tool scheduling**: permission prompts serialized, read-only tools parallelized, side-effect tools serialized
 - **Session persistence** ⭐0.5.0: unlimited archive slots, `/session` to switch anytime, tool results visible after restore. Process-level isolation — multiple instances in the same directory each get their own session slot
-- **Concurrent subagents**: three roles — `explore`/`plan`/`coder` — dispatched in parallel, streaming output visible, reports land in the conversation; per-subagent model override (`subagent` tool `model` arg or `agent.subagentModel` config — e.g. discuss with `glm-5.2`, let `deepseek-v4-flash` implement)
+- **Concurrent subagents**: five roles — `explore`/`plan`/`coder`/`eng-coder`/`eng-designer` — dispatched in parallel, streaming output visible, reports land in the conversation; per-subagent model override (`subagent` tool `model` arg or `agent.subagentModel` config — e.g. discuss with `glm-5.2`, let `deepseek-v4-flash` implement)
 - **Multi-model consultation 会诊 + 飞刀 (escalate)** ⭐0.12.30: `consult_start`/`consult_stop` run several configured models in parallel as independent read-only consultants (each with its own TUI activity card, `main_history` access to the failure trail); when every model settles, the full verdict digest is delivered automatically (R17 — consult_check retired). `escalate` flies in a stronger model for a single expert implementation run with full write access (default-async since R17 — the post-op report arrives automatically with its mutations merged). Candidate pool = `agent.consultModels` ([{ provider, model, effort? }], up to 5); budgets via `agent.consultTurns` / `agent.consultTimeoutMs`
 - **Plan Mode**: read-only exploration + design, implement after user approval
 - **AUTO mode**: `/auto` full authorization, no confirmations on long tasks
@@ -49,7 +49,8 @@ Three layers, all "query if present, skip if absent", unified hybrid retrieval:
   - IDE-native diffs — `write`/`edit` route through the editor buffer
   - Persisted sessions: list / load (history replay) / resume / delete
   - Per-session config: model / thinking / mode
-  - Setup: [docs/guides/ides.md](docs/guides/ides.md)
+  - Setup: [ACP 接入设计](../docs/cli/design/ACP-CLIENT.md)
+  - Login: `thincoder acp --login` — one interactive terminal login; ACP clients and orchestrators then drive sessions headlessly
 
 
 ## Requirements
@@ -68,7 +69,7 @@ npm install -g thincoder
 thincoder
 ```
 
-First launch walks you through a setup wizard: arrow keys to pick a provider (built-in presets or a custom endpoint) → enter API key → optionally enter an embedding key (SiliconFlow, enables vector memory search, skippable) → arrow keys to pick a model — no hand-editing config files. Adjust anytime with `/provider`, `/model`, `/config embedkey`. `chat`/`distill` also offer in-place interactive setup when no key is configured in a terminal (in pipes/CI they exit with an error and instructions).
+First launch walks you through a setup wizard: arrow keys to pick a provider (built-in presets or a custom endpoint) → enter API key → optionally enter an embedding key (SiliconFlow, enables vector memory search, skippable) → arrow keys to pick a model — no hand-editing config files. Adjust anytime with `/model` (add/remove providers, set keys, custom endpoints) or `/config embedkey`. `chat`/`distill` also offer in-place interactive setup when no key is configured in a terminal (in pipes/CI they exit with an error and instructions).
 
 You can also hand-write `~/.thincoder/config.json` (see "Configuration" below), then:
 
@@ -94,7 +95,7 @@ thincoder upgrade
 
 Running from source: replace `thincoder` above with `node bin/thincoder.mjs`.
 
-Slash commands in the TUI: `/help`, `/model` (two-level picker: first select provider, then model; `/model <provider>:<name>` switches directly), `/submodel` (subagent models per type — picker over global + explore/plan/coder/eng-coder slots, or `/submodel <type> <provider:model>` directly), `/shell` (platform-aware picker of available shells — e.g. `/shell` → pick Git Bash/pwsh, or `/shell "C:\Program Files\Git\bin\bash.exe"`, `/shell reset`; fixes win11 cmd encoding/command issues), `/provider` (add/remove providers, set keys, custom endpoints), `/think` (thinking mode toggle and reasoning effort), `/config` (view config, `/config embedkey` for the embedding key, `/config set` for parameters), `/session` (list/switch archived sessions), `/reindex` (rebuild the index), `/extract` (extract knowledge from the current session), `/restore` (restore checkpoint), `/clear`, `/exit`. High-frequency commands support abbreviations: `/h` `/x` `/m` `/p` `/t` `/c` `/n`. Typing `/` shows live matching hints in the status bar. Model picker supports search/filter — type to narrow down results.
+Slash commands in the TUI: `/help`, `/model` (two-level picker: first select provider, then model; also add/remove providers, set keys, custom endpoints; `/model <provider>:<name>` switches directly), `/submodel` (subagent models per type — picker over global + explore/plan/coder/eng-coder/eng-designer slots, or `/submodel <type> <provider:model>` directly), `/shell` (platform-aware picker of available shells — e.g. `/shell` → pick Git Bash/pwsh, or `/shell "C:\Program Files\Git\bin\bash.exe"`, `/shell reset`; fixes win11 cmd encoding/command issues), `/think` (thinking mode toggle and reasoning effort), `/config` (view config; no-argument form opens the interactive menu — e.g. default model, embedding key, proxy; `/config embedkey` sets the embedding key directly), `/session` (list/switch archived sessions), `/reindex` (rebuild the index), `/extract` (extract knowledge from the current session), `/restore` (restore checkpoint), `/clear`, `/exit`. High-frequency commands support abbreviations: `/h` `/x` `/m` `/p` `/t` `/c` `/n`. Typing `/` shows live matching hints in the status bar. Model picker supports search/filter — type to narrow down results.
 
 Configuration comes exclusively from `~/.thincoder/config.json` — no environment-variable configuration is supported.
 
@@ -102,16 +103,16 @@ Configuration comes exclusively from `~/.thincoder/config.json` — no environme
 
 ## Configuration
 
-`~/.thincoder/config.json`:
+`~/.thincoder/config.json` — strictly parsed (`JSON.parse`); the `//` annotations below are illustrative only and are not valid in the actual file:
 
-```jsonc
+```json
 {
   "providers": [
     // multiple allowed; switch with /model <name>
     {
       "name": "deepseek",
       "baseURL": "https://api.deepseek.com/v1", // any OpenAI-compatible endpoint
-      "apiKey": "sk-...", // or leave empty to use env vars
+      "apiKey": "sk-...", // or leave empty — the TUI prompts for a key when this provider is selected
       "model": "deepseek-chat",
       // optional: proactive throttling budget (match your account's rate-limit tier;
       // without it the gate is off, 429 backoff still applies).
@@ -120,7 +121,7 @@ Configuration comes exclusively from `~/.thincoder/config.json` — no environme
       // "rpm": 50,     // requests/minute
     },
   ],
-  "activeProvider": "deepseek", // currently active provider name
+  "defaultModel": "deepseek:deepseek-chat", // new-session starting model — "provider:model" (set via /config → 默认模型; /model switches the current session)
   "shell": null, // bash tool shell (win11: e.g. "C:\\Program Files\\Git\\bin\\bash.exe" or "pwsh"); null = system default — cmd on Windows (UTF-8 forced per command), /bin/sh elsewhere. TUI: /shell; a leading ~ in the path expands to the home directory
   "embedding": {
     // optional: without it, retrieval is pure FTS
@@ -131,7 +132,7 @@ Configuration comes exclusively from `~/.thincoder/config.json` — no environme
   "agent": {
     "maxTurns": 100, // tool-loop cap
     "subagentModel": null, // default subagent provider/model override: "provider:model" | provider name | model name; null = inherit parent provider. Per-call: subagent tool `model` arg
-    "subagentModels": {}, // per-type override: { "explore": "...", "plan": "...", "coder": "...", "eng-coder": "..." }; priority: tool model arg > this > subagentModel > parent provider
+    "subagentModels": {}, // per-type override: { "explore": "...", "plan": "...", "coder": "...", "eng-coder": "...", "eng-designer": "..." }; priority: tool model arg > this > subagentModel > parent provider
     "compactThreshold": 100000, // context compaction threshold (approx. tokens)
   },
   "memory": {
@@ -197,7 +198,7 @@ src/
   tui.mjs           re-export shim → src/tui/index.mjs
   tui-render.mjs    re-export shim → src/tui/render.mjs
   prompts/          prompt texts — slot-based (PROMPT-SYSTEM): persona-engineering / persona-normal /
-                    persona-{eng-coder,explore,coder,plan} + common + discipline-engineering / discipline-normal
+                    persona-{eng-coder,explore,coder,plan,eng-designer} + common + discipline-engineering / discipline-normal
                     + special modules (consult-base / advisor-design / advisor-round{1,2,3});
                     assembly = assemblePrompt (prompt-overlays.mjs): persona → common → discipline → [4] AGENTS+skills
 test/               node:test offline unit tests (npm test)
@@ -303,7 +304,7 @@ Code conventions: pure `.mjs`, no semicolons, no npm dependencies allowed (inclu
 
 ### 0.9.0 (2026-07)
 - **Checkpoint v2** — snapshot-before-destructive with full workspace copies, per-file restore (`cat` preview), auto-recover on failed apply, and escape-hatch hints when git checks fail
-- **Lifecycle Hooks** — `PreToolUse` / `PostToolUse` / `PostToolUseFailure` / `Notification` events. User-defined shell commands in config, with per-tool regex matching, timeout control, and `block`/`allow`/`notify` actions. Implemented in `src/hooks.mjs`, integrated into tool dispatch.
+- **Lifecycle Hooks** — `PreToolUse` / `PostToolUse` / `PostToolUseFailure` / `Stop` events. User-defined shell commands in config, with per-tool regex matching, timeout control, and `block`/`allow`/`notify` actions. Implemented in `thincoder-core/hooks.mjs`, integrated into tool dispatch.
 - **Built-in Skills (5)** — `pdf-create`, `xlsx-create`, `frontend-design`, `code-review`, `api-design` ship with the installation. Each is a standalone markdown instruction file using zero-dependency approaches (Chrome headless for PDF, PowerShell for Excel, etc.).
 - **Conversation message folding** — Long tool result blocks (>8 consecutive dim lines) auto-collapse to first 2 lines + "… N more lines — Enter to expand". `/fold on|off` toggles globally.
 - **Tree-shaped tasks** — `checklist` tool now supports hierarchical task IDs (`T1`, `T1.1`, `T1.2.1`) with auto-assigned numbering. `add` accepts `parent` parameter for subtree positioning. Indentation-based persistence in `checklist.md`.

package/bin/thincoder.mjs

@@ -7,7 +7,7 @@
  *   thincoder memory <sub>    Memory management: list / search / put / remove
  *   thincoder upgrade         Update to the latest version from npm
  *   thincoder completion <sh> Shell completion: bash / zsh / fish
- *   thincoder session gc    Session dir GC: --dry-run report / --confirm delete cold projects (SESSION.md §12)
+ *   thincoder session gc    Session dir GC: --dry-run report / --confirm delete cold projects (SESSION.md §6.12)
  *   thincoder acp             Agent Client Protocol server (stdio, for Zed/JetBrains/Paseo)
  *   thincoder -v              Print version
  *   thincoder --help          Print help
@@ -15,11 +15,13 @@
 
 import { readFileSync } from "node:fs"
 import { isAbsolute, join } from "node:path"
-import { runAgent } from "../src/agent.mjs"
-import { loadConfig, configPath } from "../src/config.mjs"
-import { cleanupTraces } from "../src/traces/trace-store.mjs"
-import { createMemory, syncDir } from "../src/memory.mjs"
-import { assembleAgent, teamConfig, gitAuthor, validateProvider } from "../src/cli/make-agent.mjs"
+import { runAgent } from "@thincoder/core/agent.mjs"
+import { loadConfig, configPath } from "@thincoder/core/config.mjs"
+import { cleanupTraces } from "@thincoder/core/traces/trace-store.mjs"
+import { createMemory, syncDir } from "@thincoder/core/memory.mjs"
+// 批 1 CORE-DEFECT-FIXES B3：onWait 相位值域 + 文案单源（消费面禁各自枚举——PROVIDER.md §6.20）
+import { waitStatusText } from "@thincoder/core/provider/wait-status.mjs"
+import { assembleAgent, attachManifest, teamConfig, gitAuthor, validateProvider } from "../src/cli/make-agent.mjs"
 import { memoryCommand } from "../src/cli/memory-command.mjs"
 import { setupWizard } from "../src/cli/setup-wizard.mjs"
 import { summarize, askPermission } from "../src/cli/permission.mjs"
@@ -27,6 +29,13 @@ import { distillCommand } from "../src/cli/distill-command.mjs"
 import { prepareCrashReporting, recentCrashHint, writeCrashRecord } from "../src/crash-reports.mjs"
 import { setTuiActive, restoreTerminalAfterCrash } from "../src/tui/tui-lifecycle.mjs"
 import { spawnTuiWrapped } from "../src/tui/wrapped-spawn.mjs"
+import { configurePromptInjections } from "@thincoder/core/prompt-files.mjs"
+import { CLI_PROMPT_INJECTIONS } from "../src/prompt-injections.mjs"
+
+// U2（CORE-UNIFICATION §2.6.3 专项补⑦ / §2.13.8（六））：CLI 端锚取值表——**进程入口、
+// 任何装配之前**注册一次（调用期应用 ⇒ 无导入序要求）。漏配 = 锚字面静默进模型
+// ⇒ 入口面用例（test/integration/cli-prompt-entry.test.mjs）显式覆盖本调用路径。
+configurePromptInjections(CLI_PROMPT_INJECTIONS)
 
 const [command, ...args] = process.argv.slice(2)
 // TUI-STDERR-CAPTURE（F-1）：TUI 启动（tui/无命令）默认包装——父 spawn 子 tee stderr 落盘（外部
@@ -41,6 +50,13 @@ const VERSION = JSON.parse(readFileSync(new URL("../package.json", import.meta.u
 // 期窗口）——V8 OOM/原生 fatal 自动写 report.*.json（实现批实测：目录缺失时 Node 静默不写
 // ——预建为必要动作）。失败不阻断启动（尽力面）。
 prepareCrashReporting()
+// TUI-OOM-ROOTCAUSE（CRASH-REPORTS.md §8.3 武装点——全命令同源单点）：堆遥测/看门狗——
+// 60s 采样 + 双档（70/85%）比例边缘预警（stderr + TUI 行 + 事件日志）；定时器 unref
+// （一次性命令自然退出零阻塞）；默认开 / THINCODER_HEAP_WATCH 关值集可不启。失败面全吞。
+try {
+  const { startHeapWatch } = await import("../src/heap-watch.mjs")
+  startHeapWatch()
+} catch { /* 看门狗启动失败不阻断启动（尽力面） */ }
 
 // R25（F-R25a）异常钩子升级：原"只 console.error 一行（TUI 全屏下不可见）"→ ① 落盘
 // crash-{ts}-{pid}.json ② TUI 活动态终端恢复 ③ console.error（恢复后打印才可见）
@@ -86,8 +102,10 @@ const USAGE = `thincoder - thin coding agent
 
 Usage:
   thincoder                 Launch the interactive TUI
+  thincoder tui             Launch the interactive TUI (explicit form of the default)
   thincoder chat [--auto] <prompt>   One-shot agent run (tools enabled), streams reply to stdout; --auto approves all tool calls
   thincoder acp             Agent Client Protocol server (stdio — Zed/JetBrains/Paseo drive sessions)
+  thincoder acp --login     Authenticate this machine for ACP clients (terminal auth flow), then exit
   thincoder memory list [--type=<t>]           List memory entries
   thincoder memory search <query>              Search memory
   thincoder memory put --type=<t> --title=<t> --content=<c> [--tags=<t>]
@@ -108,7 +126,7 @@ Config: ~/.thincoder/config.json — providers[] (one default model per channel)
 
 /** Unified message when no API key is configured */
 function noKeyMessage() {
-  return `No API key configured yet. Run "thincoder" to enter TUI, use /provider add and /provider key; or edit ${configPath} directly`
+  return `No API key configured yet. Run "thincoder" to enter TUI, use /model to add a provider and set its key; or edit ${configPath} directly`
 }
 
 /** Delay exit: process.exit right after fetch triggers libuv assertion on Windows/Node 24; let handles drain first */
@@ -138,7 +156,7 @@ switch (command) {
     if (crashNotice) console.error(crashNotice)
 
     const agent = await assembleAgent()
-    // SESSION.md §8 D-S4（F4）+ MODEL-MERGE-SESSION：headless 无 TUI —— 可读错误 + 退出码 1，
+    // SESSION.md §6.8 D-S4（F4）+ MODEL-MERGE-SESSION：headless 无 TUI —— 可读错误 + 退出码 1，
     // 不弹 UI、不崩溃；文案改引 defaultModel（F-6 引导 A headless 面——/config → 默认模型）
     if (agent._providerInvalid) {
       const dm = agent.config?.defaultModel
@@ -162,7 +180,7 @@ switch (command) {
       // Wizard may have configured an embedding key: attach vector search
       const fresh = loadConfig()
       if (fresh.embedding?.apiKey && agent.memory && !agent.memory.embedder) {
-        const { createEmbedder } = await import("../src/embedding.mjs")
+        const { createEmbedder } = await import("@thincoder/core/embedding.mjs")
         agent.memory.embedder = createEmbedder(fresh.embedding)
       }
     }
@@ -172,8 +190,9 @@ switch (command) {
     try {
       await runAgent(agent, prompt, {
         onToken: (text) => process.stdout.write(text),
-        onWait: ({ phase, seconds }) => {
-          console.error(phase === "gate" ? `[rate-limit] TPM throttle waiting ~${seconds}s` : `[rate-limit] 429 response, retrying in ${seconds}s`)
+        onWait: (ev) => {
+          const s = waitStatusText(ev)
+          if (s) console.error(`[rate-limit] ${s}`)
         },
         onToolCall: (name, toolArgs) => {
           console.error(`\n[tool] ${name} ${summarize(toolArgs)}`)
@@ -220,7 +239,7 @@ switch (command) {
     const config = loadConfig()
     const memory = createMemory({ dbPath: config.memory.dbPath })
     if (config.embedding?.apiKey) {
-      const { createEmbedder } = await import("../src/embedding.mjs")
+      const { createEmbedder } = await import("@thincoder/core/embedding.mjs")
       memory.embedder = createEmbedder(config.embedding)
     }
     if (config.memory.projectDir) {
@@ -241,7 +260,7 @@ switch (command) {
       break
     }
     const memory = createMemory({ dbPath: config.memory.dbPath })
-    const { ensureClone, pullTeam } = await import("../src/git/gitmem.mjs")
+    const { ensureClone, pullTeam } = await import("@thincoder/core/git/gitmem.mjs")
     try {
       const cloned = await ensureClone(team)
       if (cloned) console.log(`Cloned team repo to ${team.dir}`)
@@ -274,7 +293,7 @@ switch (command) {
     }
     const team = teamConfig(config)
     if (team) {
-      const { ensureClone } = await import("../src/git/gitmem.mjs")
+      const { ensureClone } = await import("@thincoder/core/git/gitmem.mjs")
       await ensureClone(team)
       const s = await syncDir(memory, { layer: "team", dir: team.dir })
       total.added += s.added
@@ -285,20 +304,26 @@ switch (command) {
 
   case "tui":
   case undefined: {
-    const agent = await assembleAgent()
-    // SESSION.md §8 D-S1：TUI 路径在 startTUI 前清空无效 provider——空 provider 不流入 runAgent
+    // 恢复上次的会话（同一项目目录）；provider 按保存的名字切回（用户上次可能换过模型）
+    // 2026-09-05 §6.10（R4）：恢复决策按本端记录 resumeSlot（D-2 ①②③）——manifest active
+    // 只作"无记录端"的一次性继承源，不再作本端恢复第一依据（D-6）。
+    // 2026-09-18（装配门禁批 #30 · KD-M1-13 / KD-M1-15）：resumeSlot **前移至装配之前**——
+    // 装配期模式门判据 = 会话权威值（槽优先 + config 回退）；恢复记录经 slotData 形参进
+    // 装配钩子（同进程同一调用前移 ⇒ 零新增写动作；钩子内不调 resumeSlot——非纯读）。
+    const { resumeSlot, applySession, sessionDescriptor } = await import("@thincoder/core/session.mjs")
+    // F-MI7：resumeSlot = async（探测束不阻塞事件循环）——此处必须 await
+    const { slot, data } = await resumeSlot(process.cwd())
+    const agent = await assembleAgent({ slotData: data })
+    // SESSION.md §6.8 D-S1：TUI 路径在 startTUI 前清空无效 provider——空 provider 不流入 runAgent
     // （崩溃源：chat() 缺 model → 网关 400 或 fetch("undefined/...") TypeError）
     if (agent._providerInvalid) agent.provider = null
     const config = loadConfig()
-    // 恢复上次的会话（同一项目目录）；provider 按保存的名字切回（用户上次可能换过模型）
-    // 2026-09-05 §10（R4）：恢复决策按本端记录 resumeSlot（D-2 ①②③）——manifest active
-    // 只作"无记录端"的一次性继承源，不再作本端恢复第一依据（D-6）。
-    const { resumeSlot, applySession } = await import("../src/session.mjs")
-    const { slot, data } = resumeSlot(process.cwd())
     if (data) {
       // applySession 内部已按槽复合重算 compactThreshold（auto 时）——不再需要 switched 分支
-      applySession(agent, data)
-      // SESSION.md §11.2（2026-09-08——N6 评审 🔴 修复）：process restarted 句 = 进程级
+      // TUI-OOM-ROOTCAUSE（SESSION.md §6.14）：传 slot → 绑定记录存储（身份核验 + 对账）
+      // ——人读线 = 尾窗（磁盘为准）；未传 = 模式 F（全量数组）
+      applySession(agent, data, { slot })
+      // SESSION.md §6.11（2026-09-08——N6 评审 🔴 修复）：process restarted 句 = 进程级
       // 信号——仅本启动 resume 路径设一次（真进程重启恢复盘上会话）；/session 切换与 ACP
       // 加载走同一 applySession 收敛但不设（不误报进程重启）。prepareRun 发句即清——进程
       // 内一次。resumed:yes 独立由 applySession 的恢复事件武装（data.history 非空）。
@@ -308,6 +333,10 @@ switch (command) {
     // "load → 首回合保存"窗口内并发方翻 active 导致的静默迁移）；/new 经 resetSessionState
     // 清 _slot（newSession 已认领 + 写记录，语义保留）
     agent._slot = slot
+    // M1 重估点（会话起点②——applySession 之后 · KD-M1-13 取值点②）：CLI 的工程模式权威值 =
+    // 会话槽（/eng 只写槽不写 config.json）；此时 agent.config.agent.engineering 已由
+    // applySession 按槽订正 ⇒ 与装配期判据同值 ⇒ 幂等重估（值同 ⇒ 无副作用）。
+    attachManifest(agent)
     // D-S3 优先级补全：applySession 可能已用会话中的有效 provider 修复（config 无效 + 会话有效）——
     // 修复后复验清除标记，仅当两者都无效才弹重选（validateProvider 幂等）
     if (agent._providerInvalid) validateProvider(agent)
@@ -326,7 +355,7 @@ switch (command) {
         projectDir: config.memory.projectDir ? (isAbsolute(config.memory.projectDir) ? config.memory.projectDir : join(process.cwd(), config.memory.projectDir)) : null,
         team: teamConfig(config),
         author: gitAuthor(),
-        restored: data,
+        restored: data ? sessionDescriptor(agent, data) : data,
         // R25（F-R25c）：TUI 启动显示一行"上次运行异常终止"提示（showStartup 渲染——无匹配不传）
         crashNotice: recentCrashHint() ?? undefined,
       })
@@ -373,15 +402,21 @@ switch (command) {
   }
 
   case "acp": {
-    const { runAcpServer } = await import("../src/acp.mjs")
+    const { runAcpServer, runAcpLogin } = await import("../src/acp.mjs")
+    // §11.2 改法 4：`args` 语义 = 追加到已配置的 agent 调用——`--login` = 终端认证流程，
+    // 不进 stdio 服务模式（不带该旗标时行为零变）；退出码语义 = 0 成功 / 非 0 失败。
+    if (args.includes("--login")) {
+      exitSoon(await runAcpLogin())
+      break
+    }
     await runAcpServer()
     break
   }
 
   case "session": {
-    // SESSION.md §12：会话目录 GC 手动面（F2 冷 cwd 报告/删除——VS Code 端无 shell 通道，仅 CLI）
-    const { runSessionGc } = await import("../src/session-gc.mjs")
-    process.exitCode = runSessionGc(args)
+    // SESSION.md §6.12：会话目录 GC 手动面（F2 冷 cwd 报告/删除——VS Code 端无 shell 通道，仅 CLI）
+    const { runSessionGc } = await import("@thincoder/core/session-gc.mjs")
+    process.exitCode = await runSessionGc(args)
     break
   }
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "thincoder",
-  "version": "0.12.61",
+  "version": "0.12.63",
   "description": "Thin coding agent - zero dependencies, no build step, Node.js native. Sharp code, zero bloat.",
   "keywords": [
     "ai",
@@ -19,6 +19,9 @@
   "engines": {
     "node": ">=24"
   },
+  "dependencies": {
+    "@thincoder/core": "^0.9.1"
+  },
   "files": [
     "bin/",
     "src/",
@@ -31,12 +34,10 @@
     "url": "git+https://github.com/xinbo-tech/thincoder.git"
   },
   "scripts": {
-    "test": "node test/run-fast.mjs",
+    "test": "node test/run.mjs",
     "prepublishOnly": "npm run release:check",
     "lint": "node scripts/check-syntax.mjs",
-    "release:check": "node scripts/release-check.mjs",
-    "test:full": "node test/run-full.mjs",
-    "test:integration": "node test/run-integration.mjs"
+    "release:check": "node scripts/release-check.mjs"
   },
   "author": "liwei <liwei@51marine.com> (上海新舶)",
   "license": "MIT"

package/src/acp/bridge.mjs

@@ -19,10 +19,12 @@
  * End-of-turn is NOT a notification: `session/prompt` resolves with
  * `{ stopReason: "end_turn" }` (kimi session.ts parity).
  */
-import { detectDanger, normalizeEOL, joinWithEol } from "../tools/shared.mjs"
-import { computeEditEntry, validateEditEntry, assertEditArgsExclusive, hasLineParams } from "../tools/edit-diff.mjs"
+import { detectDanger, normalizeEOL, joinWithEol } from "@thincoder/core/tools/shared.mjs"
+import { computeEditEntry, validateEditEntry, assertEditArgsExclusive, hasLineParams } from "@thincoder/core/tools/edit-diff.mjs"
 // 第 27 批 §12.3①/③：relay 前缀文法单一权威（模块直连——不自持正则副本）。
-import { parseRelayPath } from "../agent/relay-prefix.mjs"
+import { parseRelayPath } from "@thincoder/core/agent/relay-prefix.mjs"
+// 批 1 CORE-DEFECT-FIXES B3：onWait 相位值域 + 文案单源（本面仅日志——PROVIDER.md §6.20）
+import { waitStatusText } from "@thincoder/core/provider/wait-status.mjs"
 
 /** ACP ToolKind inference (schema v1 enum) — best-effort, clients render by kind. */
 function inferToolKind(name) {
@@ -52,11 +54,18 @@ function permissionToBoolean(response) {
 
 /**
  * Build the runAgent callbacks for an ACP session.
- * @param {{ sessionId: string, notify: (m, p) => void, request: (m, p, o?) => Promise<any>, log?: (s) => void }} deps
+ * @param {{ sessionId: string, notify: (m, p) => void, request: (m, p, o?) => Promise<any>, log?: (s) => void,
+ *           clientCaps?: { fs?: { readTextFile?: boolean, writeTextFile?: boolean } } }} deps
+ *   `clientCaps` = the §3.4 snapshot (taken at `session/new`); capabilities omitted by the
+ *   client are UNSUPPORTED ⇒ the fs reverse-RPC face stays local (§11.4 — never 30s 干等).
  */
-export function buildAcpCallbacks({ sessionId, notify, request, log = () => {} }) {
+export function buildAcpCallbacks({ sessionId, notify, request, log = () => {}, clientCaps = {} }) {
   const update = (sessionUpdate, extra = {}) =>
     notify("session/update", { sessionId, update: { sessionUpdate, ...extra } })
+  // §11.4 fs 能力位（默认 false——文档「MUST treat all capabilities omitted … as UNSUPPORTED」）；
+  // 判据同构 = kimi server.ts:626-636。edit 桥需读回 + 写回 ⇒ 两位皆真才路由。
+  const canReadFile = clientCaps?.fs?.readTextFile === true
+  const canWriteFile = clientCaps?.fs?.writeTextFile === true
   let toolSeq = 0
   // D15.8（TOOLS.md §15.1）：tool id FIFO 队列——并行同名工具按 call 序配对（dispatch B1
   // 已测：并行结果回调顺序 = call 顺序——T-TS8/T-TS9）。取代旧 Map 按名覆盖（后写覆盖先写
@@ -99,7 +108,10 @@ export function buildAcpCallbacks({ sessionId, notify, request, log = () => {} }
   // ——与本地通道逐字一致（NF15.6b / AC15.10：not found / occurrences / 空 old / 空 new）。
   const EDIT_ABORT_PREFIX = "edit aborted (atomic — no files written): "
 
+  // 不变量（§11.7-2）：**任何** fs/* 反向 RPC 发出前必过客户端能力位——守卫住在本函数内
+  // （防御面：路由分支先判，守卫兜底，漏判不发静默超时）。
   const readBuffer = async (p) => {
+    if (!canReadFile) throw new Error("client does not advertise fs.readTextFile")
     try {
       const read = await request("fs/read_text_file", { sessionId, path: p }, { timeoutMs: 30000 })
       return read?.text ?? read?.content ?? ""
@@ -108,6 +120,7 @@ export function buildAcpCallbacks({ sessionId, notify, request, log = () => {} }
     }
   }
   const writeBuffer = async (p, content) => {
+    if (!canWriteFile) throw new Error("client does not advertise fs.writeTextFile")
     try {
       await request("fs/write_text_file", { sessionId, path: p, content }, { timeoutMs: 30000 })
     } catch (e) {
@@ -176,12 +189,14 @@ export function buildAcpCallbacks({ sessionId, notify, request, log = () => {} }
       // Strip the subagent `[model]` metadata token ([model]<name>) — it's a
       // TUI/webview display signal, not conversation content, and must not reach ACP clients.
       if (/^\[model\]/.test(payload)) return
-      // D7 (AGENT-LOOP.md §7.2 + §19.5 round2 #6 + D-M7b): strip ⟦ev⟧ event tokens (turn/
-      // approval/done/settled/stopped/async — async = §19.5 D-M7b zero-field spawn marker)
-      // — they carry RS control characters and are a TUI display signal; structured ACP
-      // mapping (tool_call_update) is tracked separately in docs/TODO.md.
+      // D7 + 批 1 B2/D13（ACP-CLIENT.md §7.2）：剥离判据 = **形态**（`⟦ev⟧<小写名>` + RS
+      // 终止符同现），不再枚举事件名——枚举是漏项发生器（`queued` / `cancelled` 发射点在位
+      // 而白名单未跟 ⇒ 随 agent_message_chunk 泄漏给客户端）。终止符约束 load-bearing：
+      // 放宽为无终止符形态会吞真实正文（先例教训见 src/tui/render.mjs:250-252）。
+      // 事件 token 是 TUI 显示信号；结构化 ACP 映射（tool_call_update）另行跟踪
+      // （docs/TODO.md）。
       // 有意为之：控制字符协议/转义序列剥离正则（ANSI/⟦ev⟧/SGR/history 双线分隔）
-      if (/^⟦ev⟧(?:turn|approval|done|settled|stopped|async)\x1e/.test(payload)) return
+      if (/^⟦ev⟧[a-z]+\x1e/.test(payload)) return
       if (!payload) return // D6：剥后空载荷不发通知（防噪声空 chunk）
       update("agent_message_chunk", { content: { type: "text", text: payload } })
     },
@@ -193,7 +208,10 @@ export function buildAcpCallbacks({ sessionId, notify, request, log = () => {} }
       update("agent_thought_chunk", { content: { type: "text", text: payload } })
     },
     onUsage: (usage) => update("usage_update", { usage }),
-    onWait: ({ phase, seconds }) => log(`[rate-limit] ${phase} waiting ~${seconds}s`),
+    onWait: (ev) => {
+      const s = waitStatusText(ev)
+      if (s) log(`[rate-limit] ${s}`)
+    },
     onCompress: () => log("[context] auto-compacted"),
 
     onToolCall: (name, args, toolId) => {
@@ -265,17 +283,19 @@ export function buildAcpCallbacks({ sessionId, notify, request, log = () => {} }
     },
 
     /**
-     * fs reverse-RPC router (dispatch.mjs toolRouter, M2):
-     * - write            → fs/write_text_file (full content, no read-back)
+     * fs reverse-RPC router (dispatch.mjs toolRouter, M2) — 能力位 + 工具形态双判（§11.4）：
+     * - write            → fs/write_text_file (full content, no read-back) — 需 clientCaps.fs.writeTextFile
      * - edit              → fs/read_text_file → computeEditEntry（本地权威——单/数组形态）→ fs/write_text_file
      *                      （§15.1 D15.7 委派——双通道同语义；数组=原子批量——逐条目串行累积）
+     *                      — 需 readTextFile ∧ writeTextFile（读回是 edit 桥的必要前提）
      * - apply_patch       → local (unified-diff application is not routed in M2)
      * - delete, reads     → local
+     * 未宣告 ⇒ `{ handled: false }` 回落本地（零反向 RPC、零超时等待）。
      */
     toolRouter: async (name, args) => {
       const base = name.includes("/") ? name.split("/").pop() : name
       const path = pathOf(args)
-      if (base === "write" && path) {
+      if (base === "write" && path && canWriteFile) {
         if (typeof args?.content !== "string") {
           return { handled: true, result: `Error: write content must be a string (got ${typeof args?.content})` }
         }
@@ -290,7 +310,7 @@ export function buildAcpCallbacks({ sessionId, notify, request, log = () => {} }
       // 2026-09-08 D1：单形态按行号改（path + line/startLine/endLine + new_string——无
       // old_string）同样走 IDE 缓冲通道（hasLineParams 判定）——否则回落本地写盘会与
       // IDE 缓冲脱敏；editSingle → computeEditEntry 行号语义自动继承。
-      if (base === "edit" && (Array.isArray(args?.edits) || (path && typeof args?.new_string === "string" && (typeof args?.old_string === "string" || hasLineParams(args))))) {
+      if (base === "edit" && canReadFile && canWriteFile && (Array.isArray(args?.edits) || (path && typeof args?.new_string === "string" && (typeof args?.old_string === "string" || hasLineParams(args))))) {
         try {
           const text = Array.isArray(args?.edits) ? await editBatch(args) : await editSingle(path, args)
           return { handled: true, result: text }

package/src/acp/client-caps.mjs

@@ -0,0 +1,86 @@
+/**
+ * client-caps.mjs — `initialize` / `authenticate` handlers + the server-level
+ * client-capability snapshot (ACP-CLIENT.md §3.4 · §11.2 · §11.3).
+ *
+ * `initialize` is the protocol's ONLY capability-exchange point (schema:
+ * `InitializeRequest.clientCapabilities`). The snapshot written here is read once
+ * — by the auth-methods gate (this module) and the fs reverse-RPC gate
+ * (`bridge.mjs`) — and never mutated again (protocol semantics). Omitted
+ * capabilities are UNSUPPORTED (文档逐字「MUST treat all capabilities omitted in
+ * the initialize request as UNSUPPORTED」) — never optimistic defaults.
+ */
+import { ACP_ERRORS } from "./transport.mjs"
+
+/** Protocol versions this agent speaks (schema `ProtocolVersion`; v1 — §11.3 改法①). */
+export const SUPPORTED_PROTOCOL_VERSIONS = [1]
+
+/** Version negotiation (文档逐字): same version when supported, else the latest we support. */
+export function negotiateProtocolVersion(requested, supported = SUPPORTED_PROTOCOL_VERSIONS) {
+  const latest = Math.max(...supported)
+  return Number.isInteger(requested) && supported.includes(requested) ? requested : latest
+}
+
+/** `initialize.authMethods` entries (schema `AuthMethod` objects — `AuthMethodTerminal`
+ *  requires `id` + `name`). The `terminal` entry MUST be advertised **only when the client
+ *  enabled its terminal auth capability** (`clientCaps.auth.terminal === true`); otherwise
+ *  `[]`. `args` are APPENDED to the configured agent invocation (the descriptor cannot
+ *  provide a command) ⇒ the client actually launches `thincoder acp --login`. */
+export function authMethodsFor(clientCaps) {
+  if (clientCaps?.auth?.terminal !== true) return []
+  return [{
+    id: "terminal",
+    type: "terminal",
+    name: "Terminal login",
+    description: "Run `thincoder acp --login` in an interactive terminal to configure the provider API key.",
+    args: ["--login"],
+  }]
+}
+
+/**
+ * Build the initialize / authenticate handlers.
+ * @param {{ version: string, clientCaps: { current: object },
+ *           requireConfigured: () => { error?: object }, log?: (s: string) => void }} deps
+ *   `requireConfigured` = the shared credential gate (§11.5 — lives in handlers-session.mjs).
+ * @returns {Record<string, (params: object) => object>}
+ */
+export function createCapsHandlers({ version, clientCaps, requireConfigured, log = () => {} }) {
+  return {
+    initialize: (params) => {
+      const caps = params?.clientCapabilities && typeof params.clientCapabilities === "object" ? params.clientCapabilities : {}
+      clientCaps.current = caps // §3.4 — written once (initialize is the only exchange point)
+      log(`[acp] initialize — protocolVersion=${JSON.stringify(params?.protocolVersion)} auth.terminal=${caps.auth?.terminal === true}`)
+      return {
+        protocolVersion: negotiateProtocolVersion(params?.protocolVersion),
+        // 能力面如实声明（§11.3 改法③）：已实现的 loadSession + 四键 sessionCapabilities；
+        // promptCapabilities 全 false（本面不接 image/audio/embeddedContext——不虚报）；
+        // mcpCapabilities / additionalDirectories 省略 = 不支持（G7 / G4 本批不做）。
+        agentCapabilities: {
+          loadSession: true,
+          promptCapabilities: { image: false, audio: false, embeddedContext: false },
+          sessionCapabilities: { list: {}, resume: {}, delete: {}, close: {} },
+        },
+        authMethods: authMethodsFor(caps),
+        agentInfo: { name: "thincoder", version },
+      }
+    },
+
+    authenticate: (params) => {
+      // 契约形状（§11.2 改法③）：methodId 不属于**当前已宣告**的方法 → -32602；
+      // 凭据不可解析 → -32000；通过 → 空结果 `{}`（AuthenticateResponse 只有 `_meta`）。
+      // 注：契约禁止客户端以 `terminal` 调本方法——若客户端仍调且该法已宣告，本实现宽容放行。
+      const advertised = authMethodsFor(clientCaps.current)
+      const methodId = params?.methodId
+      if (!advertised.some((m) => m.id === methodId)) {
+        return {
+          error: {
+            ...ACP_ERRORS.INVALID_PARAMS,
+            message: `unknown auth method: ${String(methodId)} (advertised: ${advertised.map((m) => m.id).join(", ") || "none"})`,
+          },
+        }
+      }
+      const gate = requireConfigured()
+      if (gate.error) return gate
+      return {}
+    },
+  }
+}