cc-devflow 4.5.2 → 4.5.4

package/.claude/skills/cc-do/scripts/write-task-checkpoint.sh

@@ -9,7 +9,7 @@ set -euo pipefail
 usage() {
   cat <<'EOF'
 Usage:
-  write-task-checkpoint.sh --dir path/to/change --task T001 --status pending|running|passed|failed|skipped --summary "..." [--event context_ready] [--attempt 0] [--session session-id] [--next-action "..."]
+  write-task-checkpoint.sh --dir path/to/change --task T001 --status pending|running|passed|failed|skipped --summary "..." [--event context_ready] [--attempt 0] [--session session-id] [--next-action "..."] [--tdd-json '{"red":...}']
 EOF
 }
 
@@ -23,6 +23,7 @@ EVENT_TYPE=""
 ATTEMPT="0"
 SESSION_ID=""
 NEXT_ACTION=""
+TDD_JSON=""
 
 while [[ $# -gt 0 ]]; do
   case "$1" in
@@ -34,6 +35,7 @@ while [[ $# -gt 0 ]]; do
     --attempt) ATTEMPT="$2"; shift 2 ;;
     --session) SESSION_ID="$2"; shift 2 ;;
     --next-action) NEXT_ACTION="$2"; shift 2 ;;
+    --tdd-json) TDD_JSON="$2"; shift 2 ;;
     -h|--help) usage; exit 0 ;;
     *) echo "Unknown arg: $1" >&2; usage; exit 1 ;;
   esac
@@ -57,6 +59,15 @@ if [[ -z "$SESSION_ID" ]]; then
   SESSION_ID="${TASK_ID}-$(date -u +%s)"
 fi
 
+tdd_payload="null"
+if [[ -n "$TDD_JSON" ]]; then
+  if [[ -f "$TDD_JSON" ]]; then
+    tdd_payload="$(jq -c . "$TDD_JSON")"
+  else
+    tdd_payload="$(printf '%s' "$TDD_JSON" | jq -c .)"
+  fi
+fi
+
 jq -nc \
   --arg changeId "$change_id" \
   --arg taskId "$TASK_ID" \
@@ -66,6 +77,7 @@ jq -nc \
   --arg summary "$SUMMARY" \
   --arg timestamp "$timestamp" \
   --arg attempt "$ATTEMPT" \
+  --argjson tdd "$tdd_payload" \
   '{
     changeId: $changeId,
     taskId: $taskId,
@@ -75,7 +87,7 @@ jq -nc \
     summary: $summary,
     timestamp: $timestamp,
     attempt: ($attempt | tonumber)
-  }' > "$runtime_task_dir/checkpoint.json"
+  } + (if $tdd == null then {} else {tdd: $tdd} end)' > "$runtime_task_dir/checkpoint.json"
 
 if [[ -n "$EVENT_TYPE" || "$STATUS" == "failed" ]]; then
   jq -nc \

package/.claude/skills/cc-investigate/CHANGELOG.md

@@ -1,5 +1,29 @@
 # CC-Investigate Skill Changelog
 
+## v1.2.1 - 2026-04-29
+
+- add persistent debug session fields for active hypothesis, probes, cleanup state, and next evidence action
+- add diagnose-only and workflow-forensics modes so root-cause reports do not masquerade as completed repairs
+- update the analysis template with debug session, workflow forensics, and diagnose-only outcome sections
+
+## v1.2.0 - 2026-04-28
+
+- treat feedback loops as investigation products that must be made faster, sharper, and more deterministic before root cause freeze
+- require flaky investigations to raise reproduction rate with stress, repetition, timing-window narrowing, or differential loops instead of guessing from weak signals
+- add prevention handoff so confirmed root causes produce either a regression task, architecture finding, or explicit non-recorded reason
+
+## v1.1.6 - 2026-04-28
+
+- clarify that investigation domain language and durable decisions come from cc-devflow native sources: `devflow/specs/`, roadmap/backlog handoff, historical design/analysis, and change metadata
+- remove external context/architecture-decision files from the standard investigation contract so they are not implied as generated artifacts
+- route conflicts through capability specs, roadmap decisions, or historical design decisions instead of external decision-document language
+
+## v1.1.5 - 2026-04-28
+
+- add a feedback-loop contract so investigations record loop type, command, symptom match, runtime, determinism, failure rate, signal specificity, and sharpening plan before freezing root cause
+- require ranked candidate hypotheses before narrowing to active falsification targets, plus probe tags for cleanup-safe diagnostic instrumentation
+- add performance-regression, native domain/decision context, correct-test-seam, and evidence-request fields across the analysis, task, manifest, playbook, and investigation contract templates
+
 ## v1.1.4 - 2026-04-28
 
 - add boundary-probe, backward-trace, reference-comparison, diagnostic-instrumentation, and condition-wait investigation modes for multi-component, deep-stack, similar-path, and flaky failures

package/.claude/skills/cc-investigate/PLAYBOOK.md

@@ -12,12 +12,16 @@
 ## Core Rules
 
 1. 先复现，再猜原因。
-2. 先看最近变化，再决定是不是 regression。
-3. 先证伪假设，再冻结根因。
-4. `planning/analysis.md` 和 `planning/tasks.md` 必须足够让 `cc-do` 脱离当前会话继续工作。
-5. 调查失败三次后先重建入口，不准继续乱补。
-6. 没有 frozen root-cause contract，不准进入 repair task。
-7. 多组件、深层调用、flaky 问题必须先补边界探针、反向追踪或条件等待证据。
+2. 先把复现做成快、准、可复跑的 feedback loop。
+3. 先确认 loop 复现的是用户报告的同一个失败。
+4. 先看最近变化，再决定是不是 regression。
+5. 先证伪假设，再冻结根因。
+6. `planning/analysis.md` 和 `planning/tasks.md` 必须足够让 `cc-do` 脱离当前会话继续工作。
+7. 调查失败三次后先重建入口，不准继续乱补。
+8. 没有 frozen root-cause contract，不准进入 repair task。
+9. 多组件、深层调用、flaky 问题必须先补边界探针、反向追踪或条件等待证据。
+10. diagnose-only 只能输出根因、owner、风险和 next action，不能把未修复状态标成完成。
+11. workflow forensics 先分类 artifact / git / state / tool / permission / process failure，再决定是否进入修复。
 
 ## Iron Law
 
@@ -36,18 +40,21 @@ root-cause contract 至少包含：稳定复现或缩小后的可验证症状、
 ## Investigation Standard
 
 1. 先收集 symptom、expected、actual、repro。
-2. 先查 prior investigations、TODOS/backlog、report-card finding 和最近变更。
-3. 先沿代码路径定位触点和最近变更。
-4. 先做 pattern analysis，再形成 1-3 个可证伪假设。
-5. 每个假设都要写支持证据、反证、证伪方法、预期观察、实际观察。
-6. 只有被证据钉死的根因才能进入 repair contract。
-7. repair contract 只讲最小修复边界，不顺手发明新范围。
+2. 先构造 feedback loop：失败测试、HTTP 脚本、CLI fixture、浏览器脚本、trace replay、throwaway harness、fuzz、bisect、differential，最后才是 HITL。
+3. 记录 loop 的运行时间、确定性、失败率、症状匹配证据和 sharpen 计划。
+4. 先查 prior investigations、TODOS/backlog、report-card finding 和最近变更。
+5. 先沿代码路径定位触点和最近变更。
+6. 先做 pattern analysis，再列 3-5 个候选假设并收敛到 1-3 个 active hypotheses。
+7. 每个假设都要写支持证据、反证、证伪方法、预期观察、实际观察。
+8. 只有被证据钉死的根因才能进入 repair contract。
+9. repair contract 只讲最小修复边界，不顺手发明新范围。
 
 ## Investigation Modes
 
 | Mode | 什么时候用 | 第一动作 |
 | --- | --- | --- |
 | `reproduce-first` | 症状真实但不稳定 | 缩小复现命令 / 手动路径 |
+| `feedback-loop` | 已有复现但信号慢、松、偶然或不确定是否同一 bug | 记录 loop type、命令、runtime、determinism、failure rate 和 symptom match |
 | `diff-trace` | 昨天可用、今天坏了 | `git log --oneline -20 -- <affected-files>` |
 | `boundary-probe` | API -> service -> DB、CI -> build -> deploy 这类链路断裂 | 记录每层输入、输出、配置和状态 |
 | `backward-trace` | 错误出现在深层堆栈或坏值来源不明 | 从 immediate failure site 反追 original trigger |
@@ -56,6 +63,8 @@ root-cause contract 至少包含：稳定复现或缩小后的可验证症状、
 | `history-trace` | 同一区域反复坏 | 查历史 `analysis.md`、TODO、report-card finding |
 | `pattern-research` | 陌生框架 / 依赖 / 平台错误 | 脱敏后查通用错误类型 |
 | `contract-check` | 修复边界可能扩大 | 判定 implementation drift / missing spec truth / roadmap mismatch |
+| `diagnose-only` | 用户只要问题解释或现在不能修 | 冻结 root cause、owner、risk、next action，不生成实现完成态 |
+| `workflow-forensics` | devflow artifact、git、状态、权限或工具链断裂 | 分类 failure owner 和 rescue action，再决定 reroute |
 
 ## Pattern Analysis
 
@@ -68,9 +77,12 @@ root-cause contract 至少包含：稳定复现或缩小后的可验证症状、
 - configuration drift：本地 / CI / 生产表现不同
 - stale cache：清缓存后恢复或旧状态复现
 - resource leak：OOM、句柄增长、生命周期未释放
+- performance regression：变慢、CPU / IO / 查询耗时升高、吞吐下降
 - trust boundary drift：外部输入、LLM 输出、用户输入被当成可信
 - timing guess / flaky wait：任意 sleep / timeout / setTimeout 掩盖真实条件
 
+性能回归先建 baseline、profiler、query plan 或 bisect，不把普通日志当性能证据。
+
 ## Boundary And Trace Evidence
 
 复杂链路必须在 `analysis.md` 写清：
@@ -78,7 +90,12 @@ root-cause contract 至少包含：稳定复现或缩小后的可验证症状、
 - Boundary Probe Matrix：component boundary、input observed、output observed、config/env observed、state observed、verdict
 - Backward Trace Chain：immediate failure site、caller chain、bad value origin、original trigger、why symptom-site fix is rejected
 - Reference Comparison：similar working example、broken path、differences accepted / ruled out
-- Diagnostic Instrumentation Plan：probe location、question answered、command、expected signal、cleanup requirement
+- Diagnostic Instrumentation Plan：probe tag、probe location、question answered、command、expected signal、cleanup requirement
+- Feedback Loop Contract：loop type、command、expected / actual signal、symptom match、runtime、determinism、failure rate、sharpening plan
+- Correct Test Seam：test seam、public interface exercised、why it reaches the real trigger chain、why shallow tests are rejected
+- Persistent Debug Session：session id、active hypothesis、completed probes、cleanup state、next evidence action
+- Workflow Forensics：artifact state、git state、runtime state、tool/permission/process failure owner、rescue action
+- Diagnose-Only Outcome：root cause, owner, risk, next action, and explicit no-repair verdict
 
 这些字段不是装饰。它们的作用是证明根因位于源头，而不是报错点。
 
@@ -93,6 +110,10 @@ root-cause contract 至少包含：稳定复现或缩小后的可验证症状、
 
 命中历史时，写入 `analysis.md` 的 `Prior Investigations`，说明这次是复发、同类结构味道，还是无关历史。
 
+## Domain And Decision Context
+
+优先读取 cc-devflow 原生上下文：`devflow/specs/INDEX.md`、相关 capability specs、roadmap/backlog handoff、历史 `planning/design.md` / `planning/analysis.md`、`change-meta.json`。调查输出里的领域名、假设名、测试名应沿用项目词汇；如果调查结论违反 capability spec、roadmap decision 或历史 design decision，要显式写入 evidence chain，而不是静默覆盖既有设计决策。
+
 ## External Research Hygiene
 
 只有在本地证据不足、错误类型陌生、或像依赖 / 框架 / 平台问题时才做外部调研。
@@ -130,6 +151,7 @@ root-cause contract 至少包含：稳定复现或缩小后的可验证症状、
 - attempted evidence
 - why current entry is suspect
 - next option：continue / instrument-and-wait / human-review / reroute-cc-plan
+- evidence request：需要可复现环境、HAR、日志 dump、core dump、带时间戳录屏或临时生产探针权限
 - recommendation
 
 ## Local Kit

package/.claude/skills/cc-investigate/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: cc-investigate
-version: 1.1.4
+version: 1.2.1
 description: "Use when a bug, regression, broken task, or unexpected behavior needs root-cause investigation, reproducible evidence, and a frozen repair handoff before cc-do resumes coding."
 triggers:
   - "帮我查这个 bug"
@@ -33,12 +33,16 @@ writes:
 entry_gate:
   - "Read the current bug report, existing requirement artifacts, relevant code, tests, and recent history before forming any hypothesis."
   - "Use a FIX-<number>-<description> change key for new bug-fix investigations."
-  - "Reproduce or narrow the symptom first, then freeze the evidence chain before proposing repair tasks."
+  - "Build a runnable feedback loop, confirm it matches the reported symptom, then freeze the evidence chain before proposing repair tasks."
+  - "Record persistent debug session state: active hypothesis, probes, cleanup status, and next evidence action."
   - "Search prior investigations, TODO/backlog signals, and recent fixes in the affected area before declaring the bug novel."
   - "For multi-component, deep-stack, or flaky symptoms, record boundary probes, backward trace, or condition-wait evidence before declaring the root cause."
+  - "For performance regressions, collect a baseline or profile signal before treating logs as evidence."
   - "Do not write production code here; this stage ends with planning/analysis.md plus executable repair tasks for cc-do."
 exit_criteria:
   - "planning/analysis.md records symptom, reproduction, evidence chain, boundary probes or backward trace when applicable, pattern analysis, tested hypotheses, confirmed root cause, and repair boundary."
+  - "diagnose-only outcomes clearly stop before implementation while preserving root cause, owner, and next action."
+  - "workflow forensics classify artifact, git, state, or tool failures before repair tasks are generated."
   - "planning/tasks.md and planning/task-manifest.json are explicit enough that cc-do can repair the bug without chat memory."
   - "The honest next step is cc-do, cc-plan, or roadmap."
 reroutes:
@@ -130,6 +134,7 @@ NO REPAIR WITHOUT A FROZEN ROOT-CAUSE CONTRACT
 | 现实状态 | 先走什么路径 |
 | --- | --- |
 | 症状真实，但还没有稳定复现 | `reproduce-first`，先把现象钉死 |
+| 已有复现但信号慢 / 松 / 偶然 | `feedback-loop`，先把 pass/fail loop 做快、准、可复跑 |
 | 明显是 regression | `diff-trace`，先查最近变化 |
 | 多组件链路断裂 | `boundary-probe`，先记录每个边界的输入、输出、配置和状态 |
 | 报错点很深或坏值来源不明 | `backward-trace`，从 symptom site 一直追到 original trigger |
@@ -139,6 +144,8 @@ NO REPAIR WITHOUT A FROZEN ROOT-CAUSE CONTRACT
 | 错误类型陌生，像框架 / 依赖 / 平台问题 | `pattern-research`，先做脱敏外部调研 |
 | 同一区域反复坏 | `history-trace`，先查 prior investigations 和最近修复 |
 | 看起来像 bug，实则是未定义行为或新需求 | 直接 reroute 到 `cc-plan` |
+| 用户只要根因报告、不要求修复 | `diagnose-only`，停止在报告与 next action，不生成完成态实现任务 |
+| 失败来自 workflow / artifact / git / state 断裂 | `workflow-forensics`，先分类坏在文件、状态、工具、权限还是流程 |
 
 先说“这是什么类问题”，再说“我要怎么修”。没有分类的 debug，最后都会变成乱打补丁。
 
@@ -172,6 +179,9 @@ NO REPAIR WITHOUT A FROZEN ROOT-CAUSE CONTRACT
 
 `cc-investigate` 不写生产代码，不在这里偷跑 `cc-do`。
 
+diagnose-only 仍然写 `planning/analysis.md`，但 `planning/tasks.md` 只能包含证据交接、
+监控、人工动作或明确的 `reroute`；不能把“已经诊断”伪装成“已经修复”。
+
 ## Entry Gate
 
 1. 先确认当前对象仍然属于一个 requirement，而不是整个项目级故障。
@@ -186,35 +196,47 @@ NO REPAIR WITHOUT A FROZEN ROOT-CAUSE CONTRACT
    - 记录用户看见了什么
    - 记录预期与实际差异
    - 记录复现命令或手动路径
+   - 确认复现的是用户描述的同一个失败，而不是旁边的红灯
    - 如果上下文缺失，只问一个最关键问题，不一次性抛出问题清单
-2. **Trace reality**
+2. **Build feedback loop**
+   - 优先构造 agent 可运行的 pass/fail 信号：失败测试、curl / HTTP 脚本、CLI fixture、浏览器脚本、trace replay、throwaway harness、property / fuzz loop、bisect harness、differential loop，最后才是 HITL 脚本
+   - 记录 loop 类型、命令、运行时间、确定性、失败率、症状匹配证据和下一步 sharpen 计划
+   - loop 太慢、太宽、太 flaky 时，先优化 loop 本身；没有可信 loop，不进入 confirmed root cause
+   - 如果确实无法建 loop，写明尝试过什么，并请求 HAR、日志 dump、core dump、带时间戳录屏、可复现环境访问或临时生产探针权限
+3. **Trace reality**
    - 沿着代码路径找触点
    - 多组件系统先写 `Boundary Probe Matrix`：每个边界的输入、输出、配置 / 环境、状态和 pass/fail
    - 深层报错先写 `Backward Trace Chain`：immediate failure site、caller chain、bad value origin、original trigger
    - 查最近提交和同类改动
    - 查既有 `devflow/changes/*/planning/analysis.md`、`TODOS.md`、backlog、report-card finding
+   - 如果仓库有 `devflow/specs/`、roadmap/backlog handoff、历史 `planning/design.md` / `planning/analysis.md` 或 `change-meta.json`，把领域词汇和已冻结决策当成契约证据
    - 找现有测试和断点证据
    - 判定偏离的是 capability boundary、invariant，还是只是实现细节
-3. **Classify pattern**
-   - 判定是否属于 race condition、null propagation、state corruption、integration failure、configuration drift、stale cache、resource leak、trust boundary drift、timing guess / flaky wait
+4. **Classify pattern**
+   - 判定是否属于 race condition、null propagation、state corruption、integration failure、configuration drift、stale cache、resource leak、performance regression、trust boundary drift、timing guess / flaky wait
    - 如果有同仓库 working example，先写 `Reference Comparison`，列出 working path、broken path、差异和被接受 / 排除的假设
    - 如果错误类型陌生，先做脱敏外部调研；只搜索通用错误类型、框架 / 库名和版本，不搜索 raw secret / path / customer data
-4. **Form hypotheses**
-   - 只保留 1-3 个可被证伪的假设
-   - 每个假设都写支持证据和反证
+5. **Form hypotheses**
+   - 先列 3-5 个候选假设并排序，避免第一直觉锚定
+   - 再收敛到 1-3 个 active hypotheses 进入验证
+   - 每个假设都写支持证据、反证和优先级理由
    - 每个假设都写 `falsification method`、`expected observation`、`actual observation`
-5. **Test hypotheses**
+6. **Test hypotheses**
    - 用复现、日志、断言、最小探针验证
-   - 临时探针必须写 `Diagnostic Instrumentation Plan`：probe location、question answered、command、expected signal、cleanup requirement
+   - 临时探针必须写 `Diagnostic Instrumentation Plan`：probe tag、probe location、question answered、command、expected signal、cleanup requirement
+   - 每个 probe 只回答一个假设预测；一次只改一个变量
+   - debug 日志必须带唯一前缀，例如 `[DEBUG-FIX123-a4f2]`，进入 `cc-do` 前用前缀 grep 清理或转正
    - 三次假设都失败，就停下进入 escalation decision
-6. **Freeze repair contract**
+7. **Freeze repair contract**
    - 根因确认后，写进 `planning/analysis.md`
    - 只保留最小修复边界
+   - 写清正确测试缝隙：测试是否覆盖真实触发链；如果没有正确 seam，这本身就是需要记录的架构事实
    - 写明 affected module、allowed files、forbidden files、blast radius estimate；超过 5 个文件默认拆分或 reroute
    - 输出 `planning/tasks.md` + `planning/task-manifest.json` + `change-meta.json`
-7. **Hand off**
+8. **Hand off**
    - 下一步明确写成 `cc-do`
    - 如果 repair contract 越过当前 requirement，就 reroute 到 `cc-plan`
+   - 如果是 diagnose-only，下一步写成 human action、monitoring、backlog、`cc-plan` 或 `cc-do`，但不得标记实现完成
 
 ## Pattern Analysis
 
@@ -229,6 +251,7 @@ NO REPAIR WITHOUT A FROZEN ROOT-CAUSE CONTRACT
 | configuration drift | 本地可用、CI/生产失败 | env、feature flag、版本、路径、权限 |
 | stale cache | 清缓存后恢复、旧状态复现 | browser / CDN / Redis / build cache |
 | resource leak | OOM、句柄增长、慢性崩溃 | lifecycle、subscription、retention、cleanup |
+| performance regression | 变慢、CPU / IO / 查询耗时升高、吞吐下降 | baseline、profiler、query plan、bisect |
 | trust boundary drift | LLM / 用户输入 / 外部响应被当成可信 | validation、escaping、policy gate |
 | timing guess / flaky wait | sleep / setTimeout / timeout 增大后偶尔通过 | 真实完成条件、事件、文件、状态或队列计数 |
 
@@ -285,6 +308,39 @@ NO REPAIR WITHOUT A FROZEN ROOT-CAUSE CONTRACT
 
 探针不能变成修复。进入 `cc-do` 前，要么删除，要么明确写入 repair task 的清理 / 转正方式。
 
+## Feedback Loop Contract
+
+根因调查首先依赖一个可信 loop：
+
+- loop type: failing test / HTTP script / CLI fixture / browser script / trace replay / throwaway harness / property-fuzz / bisect / differential / HITL
+- command or manual driver
+- expected failing signal
+- actual failing signal
+- symptom match: 为什么它复现的是用户报告的同一个问题
+- runtime and determinism
+- failure rate for flaky bugs
+- sharpening plan: 如何让它更快、更准、更稳定
+
+把 loop 当成调查产品来迭代。已有 loop 但信号差时，先优化它：
+
+1. faster：缓存 setup、缩小 test scope、跳过无关启动。
+2. sharper：断言用户看见的具体症状，不用“没有崩溃”冒充匹配。
+3. more deterministic：固定时间、随机种子、filesystem、network、locale、feature flag。
+
+flaky bug 的目标不是立刻 100% 复现，而是提高复现率直到可调试。可以循环 100 次、并行触发、加压力、缩小时序窗口或做 differential loop；如果失败率仍低到不可证伪，先写 Evidence Request，不要继续猜。
+
+没有 loop 时，不能把代码阅读当成根因。只能写 `Evidence Request`：需要可复现环境、HAR、日志 dump、core dump、带时间戳录屏，或临时生产探针权限。
+
+## Correct Test Seam
+
+进入 repair handoff 前，必须说明回归测试缝隙是否正确：
+
+- test seam
+- public interface exercised
+- why this seam reaches the real trigger chain
+- why a shallower test would be false confidence
+- if no correct seam exists, record it as an architecture finding and keep repair verification tied to the original feedback loop
+
 ## Timing And Flaky Bugs
 
 遇到 flaky、sleep、timeout、重试后消失：
@@ -345,6 +401,14 @@ NO REPAIR WITHOUT A FROZEN ROOT-CAUSE CONTRACT
 2. 不能拆但仍是根因跨度，写明为什么。
 3. 如果已经变成设计 / 架构范围，reroute 到 `cc-plan`。
 
+## Prevention Handoff
+
+根因冻结后必须写一句后验判断：什么结构、测试 seam、capability invariant、operator guard 或文档会让这个 bug 更早暴露或根本不出现。
+
+- 如果答案是小范围 regression test，把它写进当前 repair task。
+- 如果答案是 seam / module / capability 边界问题，把它写成 architecture finding，并明确交给 `cc-plan` 或后续 backlog。
+- 如果答案只是流程提醒或人工记忆，不算预防；要么转成可执行 guard，要么明确不记录。
+
 ## Escalation Decision
 
 三次假设失败后，不准继续硬猜。`analysis.md` 必须写：
@@ -353,6 +417,7 @@ NO REPAIR WITHOUT A FROZEN ROOT-CAUSE CONTRACT
 - what was attempted
 - why current entry is suspect
 - next option：`continue-with-new-hypothesis` / `instrument-and-wait` / `human-review` / `reroute-cc-plan`
+- evidence request if the loop cannot be built or the environment is missing
 - recommendation
 
 ## Good Output
@@ -362,6 +427,7 @@ NO REPAIR WITHOUT A FROZEN ROOT-CAUSE CONTRACT
 - 假设不是列表装饰，而是带证伪方法和实际观察
 - 历史调查、最近改动、模式分析没有被跳过
 - 修复边界清楚到 `cc-do` 不需要二次调查
+- 正确测试缝隙写清楚，不用浅层测试制造假安全
 - `planning/tasks.md` 只包含修复任务，不夹带新需求
 - 如果应该回 `cc-plan`，理由写清楚，不假装还能继续 patch
 
@@ -377,14 +443,15 @@ NO REPAIR WITHOUT A FROZEN ROOT-CAUSE CONTRACT
 ## Working Rules
 
 1. 没有复现，不准声称找到了根因。
-2. 没有证据，不准把猜测写成结论。
-3. 先根因，再修复；先调查，再编码。
-4. `planning/tasks.md` 必须足够让 `cc-do` 在脱离当前对话后继续推进。
-5. 如果修复方案已经变成新 feature 设计，停止 debug，回 `cc-plan`。
-6. 三次假设失败后，默认说明你的调查入口错了，不准继续硬猜。
-7. 外部调研必须先脱敏，调研结论必须回到本仓库证据验证。
-8. 修复触点超过 5 个文件时，默认先拆分或 reroute，不把大重构伪装成 bug fix。
-9. 好的调查不是“找了很多可能性”，而是把错误世界缩成一个可信的 repair contract。
+2. 没有可信 feedback loop，不准把代码阅读包装成 confirmed root cause。
+3. 没有证据，不准把猜测写成结论。
+4. 先根因，再修复；先调查，再编码。
+5. `planning/tasks.md` 必须足够让 `cc-do` 在脱离当前对话后继续推进。
+6. 如果修复方案已经变成新 feature 设计，停止 debug，回 `cc-plan`。
+7. 三次假设失败后，默认说明你的调查入口错了，不准继续硬猜。
+8. 外部调研必须先脱敏，调研结论必须回到本仓库证据验证。
+9. 修复触点超过 5 个文件时，默认先拆分或 reroute，不把大重构伪装成 bug fix。
+10. 好的调查不是“找了很多可能性”，而是把错误世界缩成一个可信的 repair contract。
 
 ## Exit Criteria
 

package/.claude/skills/cc-investigate/assets/ANALYSIS_TEMPLATE.md

@@ -17,10 +17,36 @@
 - What the user saw:
 - Reproduction command / path:
 - Repro stability: `stable` | `intermittent` | `not-yet-reproduced` | `narrowed-only`
+- Matches reported symptom: `yes` | `no` | `partial` | `unknown`
+- Symptom match evidence:
 - Expected:
 - Actual:
 - Impact / blast radius:
 
+## Feedback Loop Contract
+
+- Loop type: `failing-test` | `http-script` | `cli-fixture` | `browser-script` | `trace-replay` | `throwaway-harness` | `property-fuzz` | `bisect` | `differential` | `hitl`
+- Command or manual driver:
+- Expected failing signal:
+- Actual failing signal:
+- Runtime:
+- Determinism: `deterministic` | `high-rate-flaky` | `low-rate-flaky` | `unknown`
+- Failure rate:
+- Signal specificity:
+- Sharpening plan:
+- If no loop, evidence request:
+
+## Debug Session
+
+- Session ID:
+- Started at:
+- Current mode: `reproduce-first` | `feedback-loop` | `diff-trace` | `boundary-probe` | `backward-trace` | `reference-compare` | `condition-wait` | `history-trace` | `pattern-research` | `contract-check` | `diagnose-only` | `workflow-forensics`
+- Active hypothesis:
+- Completed probes:
+- Open probes:
+- Cleanup status:
+- Next evidence action:
+
 ## Evidence Chain
 
 - Logs / stack traces:
@@ -29,6 +55,13 @@
 - Existing tests:
 - Prior investigations:
 - TODO / backlog / report-card signals:
+- Native domain / decision context:
+
+## Workflow Forensics
+
+| Failure surface | Observed state | Owner | Rescue action | Evidence |
+| --- | --- | --- | --- | --- |
+| artifact / git / runtime-state / tool / permission / process | | | | |
 
 ## Boundary Probe Matrix
 
@@ -55,9 +88,9 @@
 
 ## Diagnostic Instrumentation Plan
 
-| Probe location | Question answered | Command to run | Expected signal | Actual signal | Cleanup requirement |
-| --- | --- | --- | --- | --- | --- |
-| | | | | | |
+| Probe tag | Probe location | Question answered | Command to run | Expected signal | Actual signal | Cleanup requirement |
+| --- | --- | --- | --- | --- | --- | --- |
+| | | | | | | |
 
 ## Pattern Analysis
 
@@ -70,9 +103,16 @@
 | configuration drift | | ruled-out | |
 | stale cache | | ruled-out | |
 | resource leak | | ruled-out | |
+| performance regression | | ruled-out | |
 | trust boundary drift | | ruled-out | |
 | timing guess / flaky wait | | ruled-out | |
 
+## Candidate Hypotheses
+
+| Rank | Hypothesis | Why plausible | Prediction | Status |
+| --- | --- | --- | --- | --- |
+| 1 | | | | pending |
+
 ## Research Evidence
 
 - External research used: `yes` | `no`
@@ -94,6 +134,7 @@
 - Attempted evidence:
 - Why current entry is suspect:
 - Next option: `continue-with-new-hypothesis` | `instrument-and-wait` | `human-review` | `reroute-cc-plan`
+- Evidence request:
 - Recommendation:
 
 ## Root Cause
@@ -108,6 +149,24 @@
 - Operator handling after fix:
 - Prior history relationship: `new` | `recurring` | `same-root-cause` | `architectural-smell-candidate`
 
+## Diagnose-Only Outcome
+
+- Applies: `yes` | `no`
+- Why no repair now:
+- Root cause owner:
+- Risk if left unresolved:
+- Monitoring / follow-up evidence:
+- Next action: `human-action` | `monitor` | `backlog` | `reroute-cc-plan` | `handoff-cc-do`
+- Explicit no-repair verdict:
+
+## Correct Test Seam
+
+- Test seam:
+- Public interface exercised:
+- Why this seam reaches the real trigger chain:
+- Why a shallower test would be false confidence:
+- If no correct seam exists:
+
 ## Repair Boundary
 
 - Fix strategy:
@@ -125,6 +184,12 @@
 ## Review Gate
 
 - Repro stable:
+- Feedback loop trustworthy:
+- Symptom match confirmed:
 - Root cause confirmed:
+- Debug session cleanup complete:
+- Workflow forensics classified:
+- Diagnose-only verdict if applicable:
+- Correct test seam identified:
 - Repair scope still belongs to this requirement:
 - If not, reroute:

package/.claude/skills/cc-investigate/assets/TASKS_TEMPLATE.md

@@ -16,6 +16,8 @@
 - Execution mode: `single-path` | `parallel-ready`
 - Confirmed root cause:
 - Root-cause hypothesis:
+- Feedback loop:
+- Symptom match evidence:
 - Frozen repair boundary:
 - Boundary probes:
 - Backward trace:
@@ -28,16 +30,19 @@
 - Commands to trust:
 - Do not re-decide:
 - Parallel boundaries:
+- Correct test seam:
+- Evidence request if blocked:
 
 ## Phase 1: Reproduce And Probe Guard
 
 - [ ] T001 [TEST] Capture the failing behavior as a stable reproduction (dependsOn:none) `path/to/test`
-  Goal: 让 bug 先变成一个可复跑的失败事实。
+  Goal: 让 bug 先变成一个快、准、可复跑且匹配用户症状的失败事实。
   Files: `path/to/test`
   Read first: `analysis.md`, `tasks.md`
   Verification: `npm test -- path/to/test`
-  Evidence: failing output or reproducible log
-  Ready when: reproduction path 已稳定，analysis 已记录必要的 boundary / trace / comparison evidence
+  Evidence: failing output or reproducible log + symptom match evidence
+  Correct seam: test must exercise the real trigger chain through a public interface
+  Ready when: feedback loop 已稳定，analysis 已记录必要的 boundary / trace / comparison evidence
 
 ## Phase 2: Repair
 
@@ -47,7 +52,7 @@
   Read first: `analysis.md`, `path/to/test`
   Verification: `npm test -- path/to/test`
   Evidence: passing output + checkpoint
-  Ready when: T001 已证明问题存在，analysis 已证明根因源头
+  Ready when: T001 已证明同一个用户症状存在，analysis 已证明根因源头
 
 ## Phase 3: Verify
 

package/.claude/skills/cc-investigate/assets/TASK_MANIFEST_TEMPLATE.json

@@ -20,7 +20,7 @@
     ]
   },
   "planningMeta": {
-    "ccInvestigateSkillVersion": "1.1.4",
+    "ccInvestigateSkillVersion": "1.1.6",
     "analysisVersion": "analysis.v1",
     "approvedAt": "2026-04-17T12:00:00.000Z",
     "approvedBy": "user",
@@ -29,10 +29,24 @@
   "investigationMeta": {
     "symptomStatus": "stable",
     "reproductionPath": "npm test -- src/feature/feature.test.ts",
+    "feedbackLoop": {
+      "loopType": "failing-test",
+      "commandOrDriver": "npm test -- src/feature/feature.test.ts",
+      "expectedFailingSignal": "The test fails with the user-reported behavior",
+      "actualFailingSignal": "Observed failure output from the current repo",
+      "symptomMatchEvidence": "Failure output matches the reported symptom, not a nearby unrelated failure",
+      "runtime": "under 10s",
+      "determinism": "deterministic",
+      "failureRate": "100%",
+      "signalSpecificity": "asserts the exact broken behavior",
+      "sharpeningPlan": "Narrow setup or assertions if the loop becomes slow or broad",
+      "evidenceRequest": ""
+    },
     "patternAnalysis": {
-      "selectedPattern": "implementation drift",
+      "selectedPattern": "null propagation",
       "ruledOutPatterns": [
         "race condition",
+        "performance regression",
         "configuration drift",
         "timing guess / flaky wait"
       ],
@@ -73,6 +87,7 @@
     },
     "diagnosticInstrumentation": [
       {
+        "probeTag": "[DEBUG-FIXXXX-a4f2]",
         "probeLocation": "file:line or component boundary",
         "questionAnswered": "Which boundary first emits the invalid value?",
         "commandToRun": "npm test -- src/feature/feature.test.ts",
@@ -81,8 +96,23 @@
         "cleanupRequirement": "Remove temporary probe or convert it into a durable assertion/log"
       }
     ],
+    "candidateHypotheses": [
+      {
+        "rank": 1,
+        "statement": "Specific, testable root-cause claim",
+        "whyPlausible": "Reproduction output points to the affected contract",
+        "prediction": "The failing signal disappears when that contract is restored",
+        "status": "accepted-for-testing"
+      }
+    ],
     "priorInvestigations": [],
     "researchEvidence": [],
+    "domainDecisionContext": {
+      "contextFilesRead": [],
+      "adrFilesRead": [],
+      "vocabularyNotes": [],
+      "adrConflicts": []
+    },
     "rootCauseHypothesis": {
       "statement": "Specific, testable root-cause claim",
       "falsificationMethod": "Command, log probe, assertion, or code-path check",
@@ -112,6 +142,13 @@
       "nextOption": "cc-do",
       "recommendation": "Repair the confirmed root cause"
     },
+    "correctTestSeam": {
+      "testSeam": "public interface or end-to-end path that reaches the real trigger chain",
+      "publicInterfaceExercised": "CLI/API/UI behavior observed by callers",
+      "realTriggerChainCoverage": "The test enters through the same trigger path as the bug",
+      "whyShallowTestRejected": "A lower-level unit test would not prove the upstream contract",
+      "ifNoCorrectSeam": ""
+    },
     "repairBoundary": {
       "affectedModule": "src/feature",
       "allowedFiles": [
@@ -172,6 +209,8 @@
       ],
       "acceptance": [
         "The target bug is reproduced as a stable failure",
+        "The failing loop matches the user-reported symptom",
+        "The regression test uses the correct seam for the real trigger chain",
         "The failure output points to the confirmed root-cause path"
       ],
       "verification": [