peaks-loop 4.0.50 → 4.0.51

package/CHANGELOG.md

@@ -1,5 +1,19 @@
 # Changelog
 
+## 4.0.51 — 2026-09-16 (子代理 await 永远报不出结果 + 49 处"声称有测试守护、测试早已删除" + 9 个恒真断言 + 三个不可达导出)
+
+**Highlights**:
+
+1. **`peaks sub-agent await` 永远不可能报告任何结果。** `share-commands.ts` 把 `recordPaths` **硬编码成空数组**，而 `await-batch.ts` 在它为空时**直接短路** —— 于是这个命令每次都走空分支，静默返回。文件自己的注释就承认了这个状态（*"`recordPaths: []` case"*）。现在：新增 `resolveBatchRecords` 扫 `.peaks/_sub_agents/<sid>/dispatch-*.json` 并按 record **自己的 `batchId`** 匹配 —— 用的是 `finalize --batch` 与 `findBatchRecords` **已经在用的同一套约定**，没有另发明路径约定；`dispatchCount` 是真实计数；**零匹配 → `NO_DISPATCH_RECORDS` + exit 1**；ok 信封新增 `unreadableRecords` 与告警，**不可读的候选不再沉默**。反例对照：把 `recordPaths: []` 重新注入 → 新增的 7 个测试**全部失败**；还原 → 7/7 通过。
+
+2. **49 处在文档里声称"由某个测试守护"，而那些测试早就被删了 —— 包括守卫自己文件头里写的那次提交删掉的那个。** 引用完整性守卫原先的语料是 `CLAUDE.md` / `.peaks/PROJECT.md` / `README.md` / `.peaks/standards` —— **不含 `skills/**`**，而它的锚点正则里明明列着 `skills/`。把 `skills/**` 加进语料后，**守卫自己吐出 178 处悬空引用 / 64 文件**（此前临时扫描器估的 95 **低了近一倍**，且它归因的 `de0872b7` 只占 11 处）。修了 93 处后剩 85 处，**全部逐条处置、零白名单化**：其中 **43 处引用 21 个测试** —— 1 个删于 `457b9a87`（**正是守卫文件头自称"它存在的理由"的那次提交**）、20 个删于 `f17aa377`（*"delete 559 legacy unit tests"*）、**2 个从未在任何 commit 里存在过**；另有 6 处悬空文档互引。**每一条都逐站改写，不做批量正则替换**（"由 X 守护"与"见 X"语义不同）。**残留如实披露而非关闭**：这 49 站现在如实说明"无物钉住它"，并逐个点名了真实的覆盖缺口 —— `isDeprecatedReviewer` / `RD_DEPRECATED_REVIEWERS`、prereq 向后兼容形态、G11.5 标题锁、20-次调用节奏、handoff frontmatter（`gateEvidence` 在 `src/` 里**无人读取**）、superpowers 四条款规则、`peaks-ide` 审计日志助手。**未补测试（超出本片范围），也未放宽或删除任何一个。**
+
+3. **9 个 `supportsScope` 断言全部恒真，且唯一消费者是一个循环论证的测试。** 接口声明 `supportsScope(scope: 'project' | 'global')`，9 个 adapter 的实现是 1 个 `() => true` 加 8 个逐字相同、**在类型上必然为真**的表达式 —— 没有一个能返回 `false`。**生产零调用者**：全仓唯一消费者是一个把它"characterise"成两个 scope 都返回 true 的新测试。**删除**（接口 + 9 实现 + 那个测试）。理由：参数类型即值域、无 adapter 只支持单一 scope（9/9 `resolveSettingsFile` 双 scope 对称），且真实属性已被 `adapter-runtime-surfaces.test.ts` 钉住。
+
+4. **两个不可达导出与一处死参数。** `traeCnSubAgentDispatcher` 有导出、有测试、**没注册到任何 adapter**（8 个 adapter 分别注册了 claudeCode / codex / cursor / trae / null —— 且**没有 `trae-cn` adapter**，`IdeId` 里也没有）。它唯一可引的"支持"证据是 postinstall 的目录映射，**不是适配器注册** —— 给它建 adapter 等于**发明一条不存在的接线**，故删除。`awaitByLlmFallback` 生产零调用者（它的"向后兼容测试"只证明它自己存在 —— 与第 3 条同型的循环），且其注释里承诺的目标在 1.3 已兑现，故删除。另报出：`PollDispatchRecordsOptions.ide` **从未被 `pollDispatchRecords` 读取**（既有死表面，属三项目标之外，仅报出未修）。共 15 文件 **+61/−170**。
+
+5. **守卫的射程缺口与被裁掉的候选一起被记录。** 本片为 31 处误报新增了 5 条**带理由**的候选排除规则（省略号族、`./` 消费者根、`<…>` 模板、illustration 提示、`isDocumentRelative`），为 5 处运行时路径新增 `RUNTIME_STATE_PREFIXES`（`_runtime/` / `cron/` / `cache/`，各带理由）—— `OPTIONAL_RUNTIME_PATHS` **未增长（仍 2 条）**。**同时如实报出两个未修的射程缺口**：裸文件名引用（如 `skills-skill-md-naming.test.ts`）与围栏 ` ```bash ` 代码块对候选规则**不可见**。守卫的注入验证含**一条负对照**：同一路径写成 `mappers/…` 保持绿 —— 那正是本片唯一收窄候选集的地方，公开标注以免被读成"全都能抓"。
+
 ## 4.0.50 — 2026-09-16 (自审把 29% 从分母里排除 + 反漂移闸门报 15/15 而只跑了 1 条 + 一条"no exceptions"的红线没有任何执行者 + 五个测试套件从未运行过 + macOS 从未进 CI)
 
 **Highlights**:

package/README-en.md

@@ -140,7 +140,7 @@ Every lane opens with **one slash command**.
 
 | | |
 | --- | --- |
-| **Latest** | [![npm](https://img.shields.io/npm/v/peaks-loop?style=for-the-badge&logo=npm&logoColor=white&color=cb3837)](https://www.npmjs.com/package/peaks-loop) — 4.0.50 (2026-09-16) |
+| **Latest** | [![npm](https://img.shields.io/npm/v/peaks-loop?style=for-the-badge&logo=npm&logoColor=white&color=cb3837)](https://www.npmjs.com/package/peaks-loop) — 4.0.51 (2026-09-16) |
 | **Domains** | Code (`peaks-code`) · Content (`peaks-content`) · Project health (`peaks-doctor`) · Issue sweep (`peaks-issue-fix-orchestrator`) · Custom SOP (`peaks-sop`) · Cross-domain primitives (`peaks-solo` dispatcher · `peaks-resume` · `peaks-status` · `peaks-test` · `peaks-slice-decompose`) |
 | **Sediment pool** | `~/.peaks/` local pool · twice-clean runs auto-promote to a bee · broken runs come back for you to redefine · the bee grows with your taste |
 | **Test suite** | 285+ cases · 4 packages (peaks-loop / peaks-loop-mut / peaks-loop-shared-channel / peaks-loop-shared) · **0 timeouts** · 14 BDD caller-binding edge cases |

package/README.md

@@ -140,7 +140,7 @@ npm i -g peaks-loop
 
 | | |
 | --- | --- |
-| **最新版本** | [![npm](https://img.shields.io/npm/v/peaks-loop?style=for-the-badge&logo=npm&logoColor=white&color=cb3837)](https://www.npmjs.com/package/peaks-loop) — 4.0.50(2026-09-16) |
+| **最新版本** | [![npm](https://img.shields.io/npm/v/peaks-loop?style=for-the-badge&logo=npm&logoColor=white&color=cb3837)](https://www.npmjs.com/package/peaks-loop) — 4.0.51(2026-09-16) |
 | **覆盖域** | 代码(`peaks-code`) · 内容(`peaks-content`) · 项目健康(`peaks-doctor`) · 批量修 issue(`peaks-issue-fix-orchestrator`) · 自定义 SOP(`peaks-sop`) · 通用原语(`peaks-solo` 分诊 / `peaks-resume` 续 / `peaks-status` 看 / `peaks-test` 测 / `peaks-slice-decompose` 切片) |
 | **沉淀池** | `~/.peaks/` 本地池 · 跑两次自动晋升成 bee · 跑翻车让你重定义 · bee 跟着你的口味长 |
 | **测试套件** | 1096 cases · 4 packages (peaks-loop 1015 / runtime 39 / mut 22 / shared-channel 20) · **CI 首次全绿**(ubuntu + windows) · 14 BDD caller-binding coverage |

package/dist/cli/commands/share-commands.js

@@ -1,5 +1,5 @@
-import { realpathSync as realpathSyncNative } from 'node:fs';
-import { resolve } from 'node:path';
+import { existsSync, readdirSync, realpathSync as realpathSyncNative } from 'node:fs';
+import { join, resolve } from 'node:path';
 import { fail, getErrorMessage, ok } from 'peaks-loop-shared/result';
 import { addJsonOption, printResult } from '../cli-helpers.js';
 import { readSharedChannel, writeSharedEntry, SHARED_CHANNEL_SOFT_VALUE_WARN } from 'peaks-loop-shared-channel';
@@ -282,19 +282,39 @@ export function registerAwaitCommand(parent, io) {
             process.exitCode = 1;
             return;
         }
-        // No record-path index is kept for DAG-dispatched batches yet; the caller
-        // is expected to have a single shared record directory. We pass the empty
-        // list — the runner tracks outcomes through its own contract-store writes,
-        // so the dispatcher reaches no slot and reports no results (see the
-        // `recordPaths: []` case in tests/unit/services/dispatch/
-        // sub-agent-dispatchers.test.ts, which pins that empty shape).
-        const input = {
-            batchId: options.batch,
-            dispatchCount: 1,
-            recordPaths: [],
-            ...(timeoutMs !== undefined ? { timeoutMs } : {})
-        };
         try {
+            // Slice 2026-09-16-n1-await-reports: resolve the batch's records from the
+            // session's dispatch directory. `recordPaths` used to be hardcoded to
+            // `[]`, and `awaitBatch` short-circuits on an empty list
+            // (await-batch.ts:134) — so `await` reported zero results and exited 0
+            // for every batch, forever. A batch we cannot locate is now a failure,
+            // not a success with nothing in it.
+            const { readRecord } = await import('../../services/dispatch/dispatch-record-writer.js');
+            const scan = resolveBatchRecords({
+                projectRoot,
+                sessionId: sid,
+                batchId: options.batch,
+                readOne: readRecord
+            });
+            if (scan.recordPaths.length === 0) {
+                const unreadableNote = scan.unreadable.length > 0
+                    ? ` ${scan.unreadable.length} dispatch record(s) there could not be read, so they could not be matched to this batch: ${scan.unreadable.join(', ')}`
+                    : '';
+                printResult(io, fail('sub-agent.await', 'NO_DISPATCH_RECORDS', `No dispatch record with batchId=${options.batch} under ${scan.sessionDir}.${unreadableNote}`, {
+                    ok: false,
+                    batchId: options.batch,
+                    sessionDir: scan.sessionDir,
+                    unreadableRecords: scan.unreadable
+                }, [awaitErrorNextActions('NO_DISPATCH_RECORDS')]), asJson);
+                process.exitCode = 1;
+                return;
+            }
+            const input = {
+                batchId: options.batch,
+                dispatchCount: scan.recordPaths.length,
+                recordPaths: scan.recordPaths,
+                ...(timeoutMs !== undefined ? { timeoutMs } : {})
+            };
             const results = await dispatcher.awaitBatch(input);
             const summary = summarizeBatchResults(results);
             printResult(io, ok('sub-agent.await', {
@@ -303,15 +323,18 @@ export function registerAwaitCommand(parent, io) {
                 batchId: options.batch,
                 ide: dispatcher.label,
                 results,
-                summary
-            }, [], [
+                summary,
+                unreadableRecords: scan.unreadable
+            }, scan.unreadable.length > 0
+                ? [`${scan.unreadable.length} unreadable dispatch record(s) in this session were skipped and are NOT part of the results: ${scan.unreadable.join(', ')}`]
+                : [], [
                 // Slice 2026-09-15-s9: corrected. This used to tell users that the
                 // four non-Claude IDEs would report `awaitByLlm: <ide> 1.2 fallback`,
                 // the slice-1.2 marker that slice 1.3 replaced with a real
                 // file-polling await. The text survived because nothing tested it —
                 // no adapter produces that note any more (asserted in
-                // sub-agent-dispatchers.test.ts), and the only code that still emits
-                // it, `awaitByLlmFallback`, has no callers.
+                // sub-agent-dispatchers.test.ts), and the emitter that produced it,
+                // `awaitByLlmFallback`, has since been removed.
                 `Each non-claude-code IDE labels its own results (see the \`note\` field), so a timed-out slot is attributable to the adapter it came from.`
             ]), asJson);
         }
@@ -331,6 +354,9 @@ function awaitErrorNextActions(code) {
     if (code === 'IDE_NOT_SUPPORTED') {
         return 'Switch to claude-code, or rely on LLM-side await for non-claude-code IDEs in slice 1.3.';
     }
+    if (code === 'NO_DISPATCH_RECORDS') {
+        return 'Check --session-id / --project: records live under .peaks/_sub_agents/<sessionId>/. Use the batchId exactly as printed by the dispatch envelope.';
+    }
     return 'See error message; check that --batch matches the dispatch envelope and --timeout is a positive integer ms.';
 }
 /**
@@ -351,6 +377,47 @@ function safeRecordPath(p) {
         return p;
     }
 }
+/**
+ * Slice 2026-09-16-n1-await-reports: which on-disk records belong to a batch.
+ *
+ * Reuses the conventions already in the repo instead of inventing a new one:
+ *   - records live at `.peaks/_sub_agents/<sid>/dispatch-<rid>-<ts>.json`
+ *     (`dispatchRecordPath`, src/services/security/safe-settings-path.ts);
+ *   - a record's batch is its own `batchId` field, written by
+ *     `writeInitialDispatchRecord`. The `--batch` branch of `finalize` (below)
+ *     and `findBatchRecords` in heartbeat-watch-command.ts resolve a batch the
+ *     same way: scan the session dir, filter `dispatch-*.json`, compare field.
+ *
+ * `unreadable` lists the `dispatch-*.json` candidates whose batch could NOT be
+ * determined. The caller must not fold them into "no such batch": a record it
+ * cannot read is not evidence that the batch is empty.
+ */
+function resolveBatchRecords(input) {
+    const sessionDir = resolve(input.projectRoot, '.peaks', '_sub_agents', input.sessionId);
+    const recordPaths = [];
+    const unreadable = [];
+    if (!existsSync(sessionDir))
+        return { sessionDir, recordPaths, unreadable };
+    for (const name of readdirSync(sessionDir)) {
+        // `active-dispatches.json` (the index) and `batch-<uuid>.counter.json` are
+        // not records; neither carries a `version`, so `readRecord` would reject
+        // them as "version mismatch". Same filter as the `--batch` branch below.
+        if (!name.startsWith('dispatch-') || !name.endsWith('.json'))
+            continue;
+        const recordPath = safeRecordPath(join(sessionDir, name));
+        let batchId;
+        try {
+            batchId = input.readOne(recordPath).batchId;
+        }
+        catch {
+            unreadable.push(recordPath);
+            continue;
+        }
+        if (batchId === input.batchId)
+            recordPaths.push(recordPath);
+    }
+    return { sessionDir, recordPaths, unreadable };
+}
 export function registerFinalizeCommand(parent, io) {
     addJsonOption(parent
         .command('finalize')

package/dist/services/dispatch/sub-agent-dispatcher.d.ts

@@ -139,22 +139,13 @@ export declare const claudeCodeSubAgentDispatcher: SubAgentDispatcher;
  * stays the same; only the per-IDE wiring breaks (intentionally).
  */
 export declare const traeSubAgentDispatcher: SubAgentDispatcher;
-/**
- * Trae-CN dispatcher. Mirrors Trae's shape with a separate label so
- * the CLI's IDE detection can distinguish a Trae install region
- * (Trae-CN differs in skill install path / log location only;
- * dispatch surface is identical per slice #011 framework rule).
- *
- * Slice 1.3: real `awaitBatch` — same polling core as Trae.
- */
-export declare const traeCnSubAgentDispatcher: SubAgentDispatcher;
 /**
  * Codex (OpenAI CLI IDE) dispatcher.
  *
  * Slice #13 noted Codex's sub-agent tool name is TBD; per slice #009
  * rationale, the dispatcher mirrors Claude Code's shape so the
- * adapter contract stays uniform. Slice 1.3 promotes Codex from
- * `awaitByLlmFallback` to a real `awaitBatch` (file-based polling,
+ * adapter contract stays uniform. Slice 1.3 promotes Codex from the
+ * 1.2 marker to a real `awaitBatch` (file-based polling,
  * Codex default 45s — Codex's documented heartbeat is slightly
  * slower per slice #13 R-3).
  */
@@ -201,36 +192,26 @@ export declare class SubAgentNotSupportedError extends Error {
  */
 export declare function awaitClaudeCodeBatch(input: SubAgentAwaitBatchInput): Promise<readonly SubAgentBatchResult[]>;
 /**
- * Slice 1.3 — shared per-IDE polling core for trae / trae-cn / codex /
- * cursor. Same polling loop shape as `awaitClaudeCodeBatch`, with
- * per-IDE default timeout + note prefix. The 4 IDEs differ only in
- * (a) `defaultTimeoutMs` (Trae / Trae-CN / Cursor = 30s, Codex = 45s
- * per slice #13 R-3) and (b) the `note` label surfaced when an IDE
- * times out (so 1.4 dogfood can attribute a timeout to the right
- * IDE).
+ * Slice 1.3 — shared per-IDE polling core for trae / codex / cursor.
+ * Same polling loop shape as `awaitClaudeCodeBatch`, with per-IDE
+ * default timeout + note prefix. The 3 IDEs differ only in
+ * (a) `defaultTimeoutMs` (Trae / Cursor = 30s, Codex = 45s per slice
+ * #13 R-3) and (b) the `note` label surfaced when an IDE times out
+ * (so 1.4 dogfood can attribute a timeout to the right IDE).
  *
- * MVP rationale (per Karpathy §2 Simplicity First): the 4 IDEs
+ * MVP rationale (per Karpathy §2 Simplicity First): the 3 IDEs
  * currently share the same file-based polling transport. The only
  * per-IDE distinction is the timeout + label. Future per-IDE
  * divergence (real IPC / shell hooks) is a 1.4 dogfood concern —
  * here we keep the dispatcher interface uniform while each IDE's
- * `awaitBatch` is now a real implementation rather than the 1.2
- * `awaitByLlmFallback` marker.
+ * `awaitBatch` is a real implementation.
  */
 export interface PollDispatchRecordsOptions {
-    readonly ide: 'trae' | 'trae-cn' | 'codex' | 'cursor';
+    readonly ide: 'trae' | 'codex' | 'cursor';
     readonly defaultTimeoutMs: number;
     readonly notePrefix: string;
 }
 export declare function pollDispatchRecords(input: SubAgentAwaitBatchInput, opts: PollDispatchRecordsOptions): Promise<readonly SubAgentBatchResult[]>;
-/**
- * 1.2 fallback for trae / trae-cn / codex / cursor. Deprecated by
- * slice 1.3 — kept exported for legacy callers + back-compat tests
- * (the 1.2 marker is still a valid envelope shape; the 1.4 dogfood
- * tests can compare the marker note vs the 1.3 real note to verify
- * per-IDE attribution).
- */
-export declare function awaitByLlmFallback(input: SubAgentAwaitBatchInput, ide: string): Promise<readonly SubAgentBatchResult[]>;
 /**
  * In-process awaiter registration hook (used by `peaks sub-agent dispatch`
  * to attach a resolver; not part of the dispatcher public surface).

package/dist/services/dispatch/sub-agent-dispatcher.js

@@ -106,38 +106,13 @@ export const traeSubAgentDispatcher = {
         notePrefix: 'trae 1.3 real awaitBatch'
     })
 };
-/**
- * Trae-CN dispatcher. Mirrors Trae's shape with a separate label so
- * the CLI's IDE detection can distinguish a Trae install region
- * (Trae-CN differs in skill install path / log location only;
- * dispatch surface is identical per slice #011 framework rule).
- *
- * Slice 1.3: real `awaitBatch` — same polling core as Trae.
- */
-export const traeCnSubAgentDispatcher = {
-    label: 'trae-cn',
-    supportsRole: (role) => role.length > 0,
-    buildToolCall: ({ role, prompt, requestId }) => ({
-        name: 'Task',
-        args: {
-            subagent_type: 'general-purpose',
-            description: `${role} for rid=${requestId}`,
-            prompt,
-        },
-    }),
-    awaitBatch: async (input) => pollDispatchRecords(input, {
-        ide: 'trae-cn',
-        defaultTimeoutMs: 30_000,
-        notePrefix: 'trae-cn 1.3 real awaitBatch'
-    })
-};
 /**
  * Codex (OpenAI CLI IDE) dispatcher.
  *
  * Slice #13 noted Codex's sub-agent tool name is TBD; per slice #009
  * rationale, the dispatcher mirrors Claude Code's shape so the
- * adapter contract stays uniform. Slice 1.3 promotes Codex from
- * `awaitByLlmFallback` to a real `awaitBatch` (file-based polling,
+ * adapter contract stays uniform. Slice 1.3 promotes Codex from the
+ * 1.2 marker to a real `awaitBatch` (file-based polling,
  * Codex default 45s — Codex's documented heartbeat is slightly
  * slower per slice #13 R-3).
  */
@@ -245,14 +220,14 @@ export async function awaitClaudeCodeBatch(input) {
     // service. The back-compat envelope shape is preserved (one
     // `SubAgentBatchResult` per record path) so the S3 characterization
     // test stays green; the underlying loop is identical to the trae /
-    // trae-cn / codex / cursor wrappers below. The new typed outcome
+    // codex / cursor wrappers below. The new typed outcome
     // lives on the unified service; the S4 fail-fast test pins it.
     //
     // Slice 2026-07-30-nightshift: claude-code does NOT use a
     // per-IDE note prefix. The 1.4 dogfood contract says the done
     // note is `null` (raw outcome) and the failed note is the raw
-    // `outcome` string with no prefix. The 4 non-Claude IDEs
-    // (trae / trae-cn / codex / cursor) prefix the note with their
+    // `outcome` string with no prefix. The 3 non-Claude IDEs
+    // (trae / codex / cursor) prefix the note with their
     // per-IDE label so cross-IDE attribution is visible to the LLM.
     // Passing no `notePrefix` here keeps the legacy contract.
     const unified = await awaitBatchUnified(input.dispatchCount, input.recordPaths, input.timeoutMs, { defaultTimeoutMs: 60_000 });
@@ -299,24 +274,6 @@ export async function pollDispatchRecords(input, opts) {
     });
     return unified.results;
 }
-/**
- * 1.2 fallback for trae / trae-cn / codex / cursor. Deprecated by
- * slice 1.3 — kept exported for legacy callers + back-compat tests
- * (the 1.2 marker is still a valid envelope shape; the 1.4 dogfood
- * tests can compare the marker note vs the 1.3 real note to verify
- * per-IDE attribution).
- */
-export async function awaitByLlmFallback(input, ide) {
-    const startedAt = Date.now();
-    return input.recordPaths.map((p, i) => ({
-        dispatchIndex: i,
-        recordPath: p,
-        status: 'timeout',
-        durationMs: 0,
-        note: `awaitByLlm: ${ide} 1.2 fallback (real impl in 1.3)`
-    }));
-    void startedAt;
-}
 /**
  * In-process awaiter registration hook (used by `peaks sub-agent dispatch`
  * to attach a resolver; not part of the dispatcher public surface).

package/dist/services/ide/adapters/claude-code-adapter.js

@@ -591,7 +591,6 @@ export const CLAUDE_CODE_ADAPTER = {
             const root = scope === 'global' ? homedir() : resolve(projectRoot ?? homedir());
             return join(root, '.claude', 'settings.json');
         },
-        supportsScope: () => true,
     },
     envVar: 'CLAUDE_PROJECT_DIR',
     hookEvent: 'PreToolUse',

package/dist/services/ide/adapters/codex-adapter.js

@@ -54,8 +54,7 @@ export const CODEX_ADAPTER = {
         resolveSettingsFile: (scope, projectRoot) => {
             const root = scope === 'global' ? homedir() : resolve(projectRoot ?? homedir());
             return join(root, '.codex', 'settings.json');
-        },
-        supportsScope: (scope) => scope === 'project' || scope === 'global'
+        }
     },
     envVar: 'CODEX_PROJECT_DIR', // UNVERIFIED — see slice #13 PRD R-3; fallback path still works
     hookEvent: 'pre_tool_use', // UNVERIFIED — Codex lowercase snake_case pattern; re-verify against Codex docs

package/dist/services/ide/adapters/cursor-adapter.js

@@ -44,8 +44,7 @@ export const CURSOR_ADAPTER = {
         resolveSettingsFile: (scope, projectRoot) => {
             const root = scope === 'global' ? homedir() : resolve(projectRoot ?? homedir());
             return join(root, '.cursor', 'settings.json');
-        },
-        supportsScope: (scope) => scope === 'project' || scope === 'global'
+        }
     },
     envVar: 'CURSOR_PROJECT_DIR', // UNVERIFIED — see slice #12 PRD R-2; fallback path still works
     hookEvent: 'beforeShellExecution', // UNVERIFIED — see slice #12 PRD R-1; re-verify against Cursor's published hook schema

package/dist/services/ide/adapters/hermes-adapter.js

@@ -29,8 +29,7 @@ export const HERMES_ADAPTER = {
         resolveSettingsFile: (scope, projectRoot) => {
             const root = scope === 'global' ? homedir() : resolve(projectRoot ?? homedir());
             return join(root, '.hermes', 'settings.json');
-        },
-        supportsScope: (scope) => scope === 'project' || scope === 'global'
+        }
     },
     envVar: 'HERMES_PROJECT_DIR', // UNVERIFIED
     hookEvent: 'PreToolUse', // UNVERIFIED

package/dist/services/ide/adapters/openclaw-adapter.js

@@ -22,8 +22,7 @@ export const OPENCLAW_ADAPTER = {
         resolveSettingsFile: (scope, projectRoot) => {
             const root = scope === 'global' ? homedir() : resolve(projectRoot ?? homedir());
             return join(root, '.openclaw', 'settings.json');
-        },
-        supportsScope: (scope) => scope === 'project' || scope === 'global'
+        }
     },
     envVar: 'OPENCLAW_PROJECT_DIR', // UNVERIFIED
     hookEvent: 'PreToolUse', // UNVERIFIED

package/dist/services/ide/adapters/qoder-adapter.js

@@ -33,8 +33,7 @@ export const QODER_ADAPTER = {
         resolveSettingsFile: (scope, projectRoot) => {
             const root = scope === 'global' ? homedir() : resolve(projectRoot ?? homedir());
             return join(root, '.qoder', 'settings.json');
-        },
-        supportsScope: (scope) => scope === 'project' || scope === 'global'
+        }
     },
     envVar: 'QODER_PROJECT_DIR', // UNVERIFIED
     hookEvent: 'PreToolUse', // UNVERIFIED

package/dist/services/ide/adapters/tongyi-lingma-adapter.js

@@ -24,8 +24,7 @@ export const TONGYI_LINGMA_ADAPTER = {
         resolveSettingsFile: (scope, projectRoot) => {
             const root = scope === 'global' ? homedir() : resolve(projectRoot ?? homedir());
             return join(root, '.lingma', 'settings.json');
-        },
-        supportsScope: (scope) => scope === 'project' || scope === 'global'
+        }
     },
     envVar: 'TONGYI_LINGMA_PROJECT_DIR', // UNVERIFIED
     hookEvent: 'PreToolUse', // UNVERIFIED

package/dist/services/ide/adapters/trae-adapter.js

@@ -50,8 +50,7 @@ export const TRAE_ADAPTER = {
         resolveSettingsFile: (scope, projectRoot) => {
             const root = scope === 'global' ? homedir() : resolve(projectRoot ?? homedir());
             return join(root, '.trae', 'settings.json');
-        },
-        supportsScope: (scope) => scope === 'project' || scope === 'global'
+        }
     },
     envVar: 'TRAE_PROJECT_DIR',
     hookEvent: 'beforeToolCall', // VERIFIED against Trae 1.x fixture — slice 009-009-2026-06-07-trae-dogfood (2026-06-07); fixture at tests/fixtures/trae/trae-1x-payload.json

package/dist/services/ide/adapters/zcode-adapter.js

@@ -159,7 +159,6 @@ export const ZCODE_ADAPTER = {
             const root = scope === 'global' ? homedir() : resolve(projectRoot ?? homedir());
             return join(root, '.zcode', 'settings.json');
         },
-        supportsScope: (scope) => scope === 'project' || scope === 'global',
     },
     // UNVERIFIED — z-code 桌面应用未公开 `ZCODE_PROJECT_DIR` env var,这里
     // 用占位符,RD 阶段假设 z-code 借用了 Anthropic-compatible 协议并可能

package/dist/services/ide/ide-types.d.ts

@@ -41,8 +41,6 @@ export interface IdeSettingsLocation {
     readonly localSettingsFileName?: string;
     /** 解析出 settings.json 绝对路径 */
     resolveSettingsFile(scope: 'project' | 'global', projectRoot: string | undefined): string;
-    /** 该 IDE 是否支持此 scope(用于清晰报错) */
-    supportsScope(scope: 'project' | 'global'): boolean;
 }
 /**
  * Slim IDE adapter 描述。每 IDE 一个静态常量(无需 DI)。

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "peaks-loop",
-  "version": "4.0.50",
+  "version": "4.0.51",
   "description": "Loop Engineering CLI — workflow primitive / loop guards / evaluators / slice orchestration",
   "author": "SquabbyZ",
   "keywords": [
@@ -102,10 +102,10 @@
     "picomatch": "4.0.4",
     "yaml": "^2.9.0",
     "zod": "^4.4.3",
-    "peaks-loop-mut": "0.1.48",
-    "peaks-loop-internal-runtime": "0.0.35",
-    "peaks-loop-shared-channel": "0.0.52",
-    "peaks-loop-shared": "0.0.84"
+    "peaks-loop-internal-runtime": "0.0.36",
+    "peaks-loop-mut": "0.1.49",
+    "peaks-loop-shared": "0.0.85",
+    "peaks-loop-shared-channel": "0.0.53"
   },
   "devDependencies": {
     "@changesets/cli": "2.31.1",

package/skills/bee/peaks-perf-audit/SKILL.md

@@ -184,8 +184,8 @@ required fixes deduped by `(file, line, hint)`).
 - Schema: `.peaks/project-scan/audit-output-schema.md`
 - Service: `src/services/audit-independent/perf-audit-service.ts`
 - CLI: `src/cli/commands/perf-audit-commands.ts`
-- Tests: `tests/unit/services/audit-independent/perf-audit-service.test.ts`
-- Companion: `skills/peaks-security-audit/SKILL.md` (symmetric structure)
+- Tests: none — the unit suite for the service was deleted in `f17aa377`.
+- Companion: `skills/bee/peaks-security-audit/SKILL.md` (symmetric structure)
 
 ## Karpathy alignment
 

package/skills/bee/peaks-perf-audit/references/audit-protocol.md

@@ -170,4 +170,4 @@ section is the source of new sediment rows; the format is:
 - Service: `src/services/audit-independent/perf-audit-service.ts`
 - Template: `.peaks/project-scan/perf-template.md`
 - Schema: `.peaks/project-scan/audit-output-schema.md`
-- Companion: `skills/peaks-security-audit/references/audit-protocol.md`
+- Companion: `skills/bee/peaks-security-audit/references/audit-protocol.md`

package/skills/bee/peaks-prd/SKILL.md

@@ -147,7 +147,7 @@ PRD 模板在原 7 段基础上,增加 4 个**质量前置必填块**(user 写,A
 3. **UI 装配意图**(页面模式 / 关键交互 / 信息密度;**不画 ASCII 视觉**,视觉由选定的开源组件库自动决定)
 4. **上游基线**(fork 场景必填:当前 fork 基于哪个 tag / 偏离度 / 业务 patch 集)
 
-**反伪选择原则:** user 不参与技术决策(库/框架/语言/性能/安全/架构),只写业务/产品意图。AI 据业务场景块选库选架构。详见 `.peaks/memory/peaks-loop-prd-template-design.md`。
+**反伪选择原则:** user 不参与技术决策(库/框架/语言/性能/安全/架构),只写业务/产品意图。AI 据业务场景块选库选架构。详见 `.peaks/memory/peaks-cli-prd-template-design.md`。
 
 ## Project-scan gate (v2.11.0 D3 — read before brainstorm)
 
@@ -264,7 +264,7 @@ Use gstack as a concrete workflow reference for the product-facing parts of `Thi
 
 ## Authenticated product document workflow
 
-When the source PRD is an authenticated web document such as Feishu/Lark, use the Playwright MCP headed-browser surface rather than unauthenticated fetch tools. Chrome DevTools MCP is a secondary surface that only connects to an already-running Chrome (`--remote-debugging-port=9222`); it does not launch a browser on its own. The canonical browser workflow lives in `peaks-code/references/browser-workflow.md`; the rules below are the PRD-specific application.
+When the source PRD is an authenticated web document such as Feishu/Lark, use the Playwright MCP headed-browser surface rather than unauthenticated fetch tools. Chrome DevTools MCP is a secondary surface that only connects to an already-running Chrome (`--remote-debugging-port=9222`); it does not launch a browser on its own. The canonical browser workflow lives in `../../peaks-code/references/browser-workflow.md`; the rules below are the PRD-specific application.
 
 1. Confirm Playwright MCP is installed: check the LLM tool list for any Playwright MCP entry in the LLM tool list. If absent, the user installs via `claude mcp add playwright -- npx @playwright/mcp@latest` (Claude Code) or the IDE-native install command. Do not hand-edit `.claude/settings.json`.
 2. Before navigation, verify the user-provided document URL uses `https:` and belongs to an approved Feishu/Lark tenant domain such as `*.feishu.cn`, `*.larksuite.com`, `*.larksuite.com.cn`, or a project-configured tenant. Reject `file:`, `data:`, `javascript:`, `http:`, localhost, loopback, link-local, private IP, and raw IP hosts unless the user explicitly approves a controlled local test target.
@@ -294,7 +294,7 @@ Summarize business context only when it changes implementation priority, scope,
 
 When the user explicitly says the target is a frontend project, transform the product document into frontend implementation inputs before RD starts:
 
-> **v2.15.0+ 校准:** UI 视觉 = 开源组件库装配(shadcn / Radix / antd / mui),不是原创设计。prd 不画 ASCII 视觉,只描述"页面模式 + 关键交互 + 信息密度"。AI 据此选组件库 + 装配。详见 `.peaks/memory/peaks-loop-prd-template-design.md`。
+> **v2.15.0+ 校准:** UI 视觉 = 开源组件库装配(shadcn / Radix / antd / mui),不是原创设计。prd 不画 ASCII 视觉,只描述"页面模式 + 关键交互 + 信息密度"。AI 据此选组件库 + 装配。详见 `.peaks/memory/peaks-cli-prd-template-design.md`。
 
 1. identify target pages, routes, components, forms, tables, modals, empty/loading/error states, permissions, data dependencies, edge cases, and affected user flows;
 2. separate frontend-only work from API/backend联调 assumptions;

package/skills/bee/peaks-prd/references/prd-for-multi-pass.md

@@ -46,4 +46,4 @@ Single-service feature → single-pass PRD. Do not force a Pass 2 layer if the w
 - Pass 2 ACs each name a file path under one of the Pass 1 services.
 - A `peaks slice audit-prd --rid <rid>` (when present) returns zero cross-cutting flags.
 
-See `../peaks-slice-decompose/SKILL.md` for the envelope contract and `../peaks-rd/references/reading-v2-slice-results.md` for the RD-side read.
+See `../../../peaks-slice-decompose/SKILL.md` for the envelope contract and `../../peaks-rd/references/reading-v2-slice-results.md` for the RD-side read.

package/skills/bee/peaks-prd/references/workflow.md

@@ -15,7 +15,7 @@ When the product source is an authenticated Feishu/Lark/wiki document:
 7. Do not persist login URLs, redirect URLs, cookies, request or response headers, session tokens, tokens, storage state, QR payloads, raw browser state, raw network logs, browser traces, or screenshots/logs containing PII or SSO/MFA material into artifacts; redact sensitive evidence before writing `.peaks` outputs.
 8. If access remains blocked, record only a redacted document identifier, a sanitized state category such as `login-required`, `mfa-required`, or `access-denied`, and the exact user action needed.
 
-Canonical browser workflow: `../../peaks-code/references/browser-workflow.md`.
+Canonical browser workflow: `../../../peaks-code/references/browser-workflow.md`.
 
 ## Implementation-oriented analysis
 

package/skills/bee/peaks-qa/SKILL.md

@@ -47,7 +47,7 @@ denied again after a long pause. That is the gate resetting, not you regressing.
 
 > **Read once at the top of this file; the rest of the skill is written against it.**
 
-The `.peaks/` workspace is partitioned by a **single scope axis** (session-id, at `.peaks/_runtime/<sessionId>/...`) with a nested **sub-agent axis** under `.peaks/_sub_agents/<sessionId>/...`. Use `<sessionId>` placeholders (NEVER bare `<sid>`). The peaks-loop change-id axis was removed in slice `2026-06-29-change-id-root-removal`; reviewable artifacts now live under `.peaks/_runtime/<sessionId>/<role>/...` only. OpenSpec's independent `openspec/changes/<change-id>/` vocabulary (L4) is preserved untouched. CLI mapping: session-id → `peaks session *`; sub-agent → `peaks sub-agent *`. Regression test `tests/unit/skills/skills-skill-md-naming.test.ts` enforces (a) zero bare `<sid>`, (b) every `.peaks/_runtime/<X>/` has an axis label, (c) this callout is present.
+The `.peaks/` workspace is partitioned by a **single scope axis** (session-id, at `.peaks/_runtime/<sessionId>/...`) with a nested **sub-agent axis** under `.peaks/_sub_agents/<sessionId>/...`. Use `<sessionId>` placeholders (NEVER bare `<sid>`). The peaks-loop change-id axis was removed in slice `2026-06-29-change-id-root-removal`; reviewable artifacts now live under `.peaks/_runtime/<sessionId>/<role>/...` only. OpenSpec's independent `openspec/changes/<change-id>/` vocabulary (L4) is preserved untouched. CLI mapping: session-id → `peaks session *`; sub-agent → `peaks sub-agent *`. Normative by documentation only — the guard that once asserted (a) zero bare `<sid>`, (b) an axis label on every runtime path, and (c) this callout's presence was deleted in `457b9a87`. Comply by convention.
 
 ## peaks-context auto-build (v3.0)
 
@@ -92,7 +92,7 @@ The canonical scope dir for this request is provided as `envelope.data.scopeDir`
 
 When this skill is launched as a sub-agent via `peaks sub-agent dispatch <role>` (then the LLM executes the returned toolCall) from `peaks-code`, the following sections of THIS skill are **suspended** for the sub-agent run: Session id, Skill presence, Workspace initialization, Mode selection, Statusline install. The sub-agent must NOT call `peaks request init` (Code already initialised the slot), and must write `.peaks/_runtime/<sessionId>/qa/test-cases/<rid>.md` with test cases that link to PRD acceptance items. Return only a compact JSON envelope.
 
-> **v2.15.0+ 校准:** 每个 slice 完成,user 必介入做**业务审阅**(4-5 项业务/产品清单:业务流程 / 需求覆盖 / 边界 case / UI 装配 / 能合入下版吗),**不是技术审阅**。业务审过 → 进 final;业务不通过 → 返工。详见 `.peaks/memory/peaks-loop-slice-review-and-qa-perspective.md`。
+> **v2.15.0+ 校准:** 每个 slice 完成,user 必介入做**业务审阅**(4-5 项业务/产品清单:业务流程 / 需求覆盖 / 边界 case / UI 装配 / 能合入下版吗),**不是技术审阅**。业务审过 → 进 final;业务不通过 → 返工。详见 `.peaks/memory/peaks-cli-slice-review-and-qa-perspective.md`。
 
 → see `references/qa-sub-agent-dispatch.md` for the full contract + hard prohibitions.
 
@@ -106,7 +106,7 @@ Project-level security + perf plans live at `.peaks/_runtime/<sessionId>/qa/secu
 
 When peaks-qa is the **main loop** (i.e. it is the active skill and is about to run its own sub-agent dispatch, rather than being a sub-agent itself), it fans out only the **business verification** sub-agent: `qa-business`. Security and performance review are **NOT** peaks-qa's responsibility in v2.11.0 — they are owned by peaks-rd's 4-way audit fan-out (code-review + security-review + perf-baseline + karpathy-review) and the rd-side evidence files (`audit/security-<rid>.md`, `audit/perf-<rid>.md`). peaks-qa reads those files by reference; it does NOT re-do them.
 
-> **v2.15.0+ 校准:** `qa-business` 只跑业务/产品视角的 6 项验收清单(业务流程 / 需求覆盖 / 边界 case / UI 装配 / 异常态语调 / 能上线吗),**不跑技术指标**(覆盖率 / 性能 / 安全)。技术指标由 RD 4-way fan-out 自决,QA 只读 `audit/security-<rid>.md` + `audit/perf-<rid>.md`。详见 `.peaks/memory/peaks-loop-slice-review-and-qa-perspective.md`。
+> **v2.15.0+ 校准:** `qa-business` 只跑业务/产品视角的 6 项验收清单(业务流程 / 需求覆盖 / 边界 case / UI 装配 / 异常态语调 / 能上线吗),**不跑技术指标**(覆盖率 / 性能 / 安全)。技术指标由 RD 4-way fan-out 自决,QA 只读 `audit/security-<rid>.md` + `audit/perf-<rid>.md`。详见 `.peaks/memory/peaks-cli-slice-review-and-qa-perspective.md`。
 
 If the PRD or project warrants it, subdivide `qa-business` further into roles like `qa-business-api` / `qa-business-frontend` / `qa-business-regression`. Subdivision must stay ≤ 2 levels deep (RL-4).
 
@@ -182,7 +182,7 @@ QA must generate test cases, not merely inspect existing ones. Every QA invocati
 
 **Pre-drafted test cases (slice 004 optimization):** when peaks-rd's 4-way parallel fan-out ran a `qa-test-cases-writer` sub-agent, the test plan is pre-drafted at `.peaks/_runtime/<sessionId>/qa/test-cases/<rid>.md` and shipped through the rd:qa-handoff gate. QA main loop is aware of this and treats the pre-drafted file as the canonical starting point. **Missing** the pre-drafted file (sub-agent failed, or the slice was a config/docs/chore that did not fan out) → QA drafts it inline as before, falling back to the standard generation flow.
 
-> **v2.15.0+ 校准:** 业务验证层(user 必审) = 4-5 项业务/产品清单。技术验证层(AI 自决) = 覆盖率 / P99 / 安全扫描 / 自动化。两层解耦:技术不过 → AI 内部重跑;业务不过 → user 反馈修。user **不看**技术指标。详见 `.peaks/memory/peaks-loop-slice-review-and-qa-perspective.md` G5。
+> **v2.15.0+ 校准:** 业务验证层(user 必审) = 4-5 项业务/产品清单。技术验证层(AI 自决) = 覆盖率 / P99 / 安全扫描 / 自动化。两层解耦:技术不过 → AI 内部重跑;业务不过 → user 反馈修。user **不看**技术指标。详见 `.peaks/memory/peaks-cli-slice-review-and-qa-perspective.md` G5。
 
 → see `references/test-case-generation.md` for the full format + acceptance-linkage contract.
 
@@ -198,7 +198,7 @@ QA cannot pass a change until the report contains evidence for every applicable
 
 If Playwright MCP is unavailable, the LLM checks its own tool list for the Playwright MCP server entry; if absent, the LLM tells the user the install command (`claude mcp add playwright -- npx @playwright/mcp@latest` for Claude Code) and marks the gate blocked with the missing capability. Screenshots, logs, manual steps, or other tools must not substitute for the mandatory frontend browser gate. Do not silently downgrade frontend validation to API-only testing.
 
-> **v2.15.0+ 校准:** 存量项目无 UT 兜底,QA 验证必须有"轻量回归"(G14):5-10 分钟跑 10 条关键路径(关键路径来源:prd 业务场景块 / 老板强调的流程 / 历史事故 / G13 影响面扫描),**不跑完整 E2E 1-2 小时**。上线后必须走"观察期"(G15):灰度 → 监控 → 反馈聚合 → 紧急修复 → 修复回灌关键路径(防下次再犯)。详见 `.peaks/memory/peaks-loop-fast-iteration-quality-loop.md`。
+> **v2.15.0+ 校准:** 存量项目无 UT 兜底,QA 验证必须有"轻量回归"(G14):5-10 分钟跑 10 条关键路径(关键路径来源:prd 业务场景块 / 老板强调的流程 / 历史事故 / G13 影响面扫描),**不跑完整 E2E 1-2 小时**。上线后必须走"观察期"(G15):灰度 → 监控 → 反馈聚合 → 紧急修复 → 修复回灌关键路径(防下次再犯)。详见 `.peaks/memory/peaks-cli-fast-iteration-quality-loop.md`。
 
 ## Local intermediate artifacts
 
@@ -297,4 +297,4 @@ QA contracts to assert on the L2 surface (a minimal acceptance test set):
 
 ## Sub-role detached mode (Phase C, slice 2026-08-10)
 
-QA sub-roles (`qa-business`, `qa-perf`, `qa-security`, `qa-business-api`, `qa-business-frontend`, `qa-business-regression`) accept `--mode detached --vendor <vendor>` for parallel test-case execution. Detached mode spawns a real OS process via `peaks sub-agent dispatch --mode detached`, isolated from the orchestrator session. Use detached mode when the test-case writer's expected runtime exceeds 60s OR it processes ≥ 20 source files. The default remains `in-process` (backward compat; existing 106+ dispatch tests untouched). When `--mode detached` is used, the child QA agent receives the `<peaks-auto-compact>` marker (corrected 2026-09-13, rid `2026-09-13-defects-e`): it asks the child to persist durable state to `.peaks/_runtime/<sid>/detached/<rid>/compact/<n>.json` before its own harness compacts the session — the child cannot compact itself, and nothing is spliced back into its prompt. See `peaks-code/references/sub-agent-dispatch.md` §"Detached Mode" for the full contract.
+QA sub-roles (`qa-business`, `qa-perf`, `qa-security`, `qa-business-api`, `qa-business-frontend`, `qa-business-regression`) accept `--mode detached --vendor <vendor>` for parallel test-case execution. Detached mode spawns a real OS process via `peaks sub-agent dispatch --mode detached`, isolated from the orchestrator session. Use detached mode when the test-case writer's expected runtime exceeds 60s OR it processes ≥ 20 source files. The default remains `in-process` (backward compat; existing 106+ dispatch tests untouched). When `--mode detached` is used, the child QA agent receives the `<peaks-auto-compact>` marker (corrected 2026-09-13, rid `2026-09-13-defects-e`): it asks the child to persist durable state to `.peaks/_runtime/<sid>/detached/<rid>/compact/<n>.json` before its own harness compacts the session — the child cannot compact itself, and nothing is spliced back into its prompt. See `skills/peaks-code/references/sub-agent-dispatch.md` §"Detached Mode" for the full contract.

package/skills/bee/peaks-qa/references/external-capability-guidance.md

@@ -5,5 +5,5 @@
 - Playwright MCP is the required path for controlled headed browser and E2E validation (it launches a headed browser on demand). The LLM runtime exposes the Playwright tools under its own server-and-tool namespace (the Playwright MCP); QA invokes them by name from the LLM's tool list. (peaks-loop no longer auto-installs MCPs as of slice #016; the user runs `claude mcp add playwright -- npx @playwright/mcp@latest` themselves when the tool list is empty.)
 - Chrome DevTools MCP is an optional secondary surface for CDP inspection (console, network, performance) of an already-running Chrome started with `--remote-debugging-port=9222`; it does NOT launch a browser on its own. The LLM invokes Chrome DevTools MCP tools directly when present in the tool list.
 - Agent Browser can support browser walkthroughs, but never submit forms, purchase, delete, or mutate authenticated state without explicit confirmation.
-- Canonical browser workflow (URL allow-list, login handoff, sanitization rules, tool mapping): `peaks-code/references/browser-workflow.md`.
+- Canonical browser workflow (URL allow-list, login handoff, sanitization rules, tool mapping): `skills/peaks-code/references/browser-workflow.md`.
 - If Playwright MCP is not installed and the user does not authorize installation, mark frontend browser validation blocked; screenshots, logs, manual steps, or other tools must not substitute for the mandatory headed browser gate.