@fitlab-ai/agent-infra 0.7.4 → 0.7.6

package/templates/.agents/skills/create-release-note/SKILL.zh-CN.md

@@ -160,7 +160,21 @@ git log v<prev-version>..v<version> \
 
 ### 9. 发布 Release notes（如确认）
 
-按 `.agents/rules/release-commands.md` 的「发布 Release notes」命令执行（写入已由 release 工作流自动创建/发布的 Release；不存在时兜底创建）。
+9.1 把生成的 notes 写入**工作树之外**的临时文件，避免在仓库残留未提交产物（不要写入 `.agents/workspace/` 或任何受版本控制的目录）：
+
+```bash
+NOTES_FILE="$(mktemp "${TMPDIR:-/tmp}/agent-infra-release-notes.XXXXXX")"
+```
+
+把 notes 内容写入 `$NOTES_FILE`。
+
+9.2 按 `.agents/rules/release-commands.md` 的「发布 Release notes」命令执行（命令中的 `{notes-file}` 用 `$NOTES_FILE`；写入已由 release 工作流自动创建/发布的 Release，不存在时兜底创建）。
+
+9.3 无论发布成功或失败，都删除临时文件：
+
+```bash
+rm -f "$NOTES_FILE"
+```
 
 输出：
 ```
@@ -179,6 +193,7 @@ Release notes 已更新。
 2. **标签必须存在**：先执行 release 技能创建标签
 3. **Release 已自动发布**：`v{version}` 的 Release 由 release 工作流自动创建并发布（给 Homebrew bottle 提供上传落点）；本技能往该 Release 写入/刷新 notes
 4. **分类准确性**：自动分类基于标题/scope/文件；复杂的 PR 可能需要手动调整
+5. **不留残留产物**：notes 一律写入工作树之外的临时文件（`mktemp`）并在发布后删除，禁止写入仓库目录
 
 ## 错误处理
 

package/templates/.agents/skills/create-task/SKILL.en.md

@@ -26,6 +26,17 @@ Version stamp rule: when creating or updating `task.md` frontmatter, read `.agen
 
 > If `{task-id}` matches `^[#]?[0-9]+$` (bare numeric or `#`-prefixed), follow the "SKILL parameter resolver" section of `.agents/rules/task-short-id.md`; treat `{task-id}` as the resolved full `TASK-YYYYMMDD-HHMMSS` form for every downstream command.
 
+## Step Start: Capture the Start Time
+
+This skill **creates** task.md, so there is no file to write at the start. Capture `started_at` in memory before running (`date "+%Y-%m-%d %H:%M:%S%:z"`); when writing the Activity Log at the end, **append both lines at once** — the started line uses `started_at`, the done line uses the completion time, both sharing the base action (started line action gets a ` [started]` suffix, note `started`):
+
+```
+- {started_at} — **Create Task [started]** by {agent} — started
+- {done_at} — **Create Task** by {agent} — {completion summary}
+```
+
+`ai task log` pairs the two by base action onto one row (in progress → done). See the "Activity Log started / done dual-marker convention" in `.agents/rules/task-management.md`.
+
 ## Steps
 
 ### 1. Parse the User Description

package/templates/.agents/skills/create-task/SKILL.zh-CN.md

@@ -26,6 +26,17 @@ description: "根据自然语言描述创建任务"
 
 > 如果 `{task-id}` 入参匹配 `^[#]?[0-9]+$`（裸数字或带 `#` 前缀），先读取 `.agents/rules/task-short-id.md` 的「SKILL 入参解析」段执行解析；后续命令视 `{task-id}` 为解析后的全长 `TASK-YYYYMMDD-HHMMSS` 形式。
 
+## 步骤开始：记录开始时间
+
+本技能会**创建** task.md，开始时尚无文件可写。先在内存记录开始时间 `started_at`（`date "+%Y-%m-%d %H:%M:%S%:z"`）；在最后写活动日志时**一次性补两条**——started 行用 `started_at`、done 行用完成时间，二者同基名（started 行 action 加 ` [started]` 后缀、note 用 `started`）：
+
+```
+- {started_at} — **Create Task [started]** by {agent} — started
+- {done_at} — **Create Task** by {agent} — {完成说明}
+```
+
+`ai task log` 会按基名把两条配对成一行（进行中 → 已完成）。约定见 `.agents/rules/task-management.md` 的「Activity Log started / done 双标记约定」。
+
 ## 执行步骤
 ### 1. 解析用户描述
 
@@ -172,7 +183,7 @@ node .agents/scripts/validate-artifact.js gate create-task .agents/workspace/act
 
 下一步 - 执行需求分析：
   - Claude Code / OpenCode：/analyze-task {task-ref}
-  - Gemini CLI：/{{project}}:analyze-task {task-ref}
+  - Gemini CLI：/agent-infra:analyze-task {task-ref}
   - Codex CLI：$analyze-task {task-ref}
 ```
 
@@ -191,7 +202,7 @@ node .agents/scripts/validate-artifact.js gate create-task .agents/workspace/act
 
 下一步 - 执行需求分析：
   - Claude Code / OpenCode：/analyze-task {task-ref}
-  - Gemini CLI：/{{project}}:analyze-task {task-ref}
+  - Gemini CLI：/agent-infra:analyze-task {task-ref}
   - Codex CLI：$analyze-task {task-ref}
 ```
 
@@ -215,7 +226,7 @@ Issue 创建失败：
 
 下一步 - 执行需求分析：
   - Claude Code / OpenCode：/analyze-task {task-ref}
-  - Gemini CLI：/{{project}}:analyze-task {task-ref}
+  - Gemini CLI：/agent-infra:analyze-task {task-ref}
   - Codex CLI：$analyze-task {task-ref}
 
 后续如需平台同步：修复认证/网络/模板问题后，可按 `.agents/rules/create-issue.md` 对当前任务手动执行一次 Issue 创建；或手动创建/查找 Issue，并把 `issue_number` 写入 task.md，后续技能会接管级联同步。

package/templates/.agents/skills/import-codescan/SKILL.en.md

@@ -17,6 +17,17 @@ Import the specified Code Scanning (CodeQL) alert and create a remediation task.
 
 > If `{task-id}` matches `^[#]?[0-9]+$` (bare numeric or `#`-prefixed), follow the "SKILL parameter resolver" section of `.agents/rules/task-short-id.md`; treat `{task-id}` as the resolved full `TASK-YYYYMMDD-HHMMSS` form for every downstream command.
 
+## Step Start: Capture the Start Time
+
+This skill **creates** task.md, so there is no file to write at the start. Capture `started_at` in memory before running (`date "+%Y-%m-%d %H:%M:%S%:z"`); when writing the Activity Log at the end, **append both lines at once** — the started line uses `started_at`, the done line uses the completion time, both sharing the base action (started line action gets a ` [started]` suffix, note `started`):
+
+```
+- {started_at} — **Import Codescan [started]** by {agent} — started
+- {done_at} — **Import Codescan** by {agent} — {completion summary}
+```
+
+`ai task log` pairs the two by base action onto one row (in progress → done). See the "Activity Log started / done dual-marker convention" in `.agents/rules/task-management.md`.
+
 ## Execution Flow
 
 ### 1. Retrieve Alert Information

package/templates/.agents/skills/import-codescan/SKILL.zh-CN.md

@@ -17,6 +17,17 @@ description: "导入 Code Scanning 告警并创建修复任务"
 
 > 如果 `{task-id}` 入参匹配 `^[#]?[0-9]+$`（裸数字或带 `#` 前缀），先读取 `.agents/rules/task-short-id.md` 的「SKILL 入参解析」段执行解析；后续命令视 `{task-id}` 为解析后的全长 `TASK-YYYYMMDD-HHMMSS` 形式。
 
+## 步骤开始：记录开始时间
+
+本技能会**创建** task.md，开始时尚无文件可写。先在内存记录开始时间 `started_at`（`date "+%Y-%m-%d %H:%M:%S%:z"`）；在最后写活动日志时**一次性补两条**——started 行用 `started_at`、done 行用完成时间，二者同基名（started 行 action 加 ` [started]` 后缀、note 用 `started`）：
+
+```
+- {started_at} — **Import Codescan [started]** by {agent} — started
+- {done_at} — **Import Codescan** by {agent} — {完成说明}
+```
+
+`ai task log` 会按基名把两条配对成一行（进行中 → 已完成）。约定见 `.agents/rules/task-management.md` 的「Activity Log started / done 双标记约定」。
+
 ## 执行流程
 
 ### 1. 获取告警信息

package/templates/.agents/skills/import-dependabot/SKILL.en.md

@@ -17,6 +17,17 @@ Import the specified Dependabot security alert and create a remediation task.
 
 > If `{task-id}` matches `^[#]?[0-9]+$` (bare numeric or `#`-prefixed), follow the "SKILL parameter resolver" section of `.agents/rules/task-short-id.md`; treat `{task-id}` as the resolved full `TASK-YYYYMMDD-HHMMSS` form for every downstream command.
 
+## Step Start: Capture the Start Time
+
+This skill **creates** task.md, so there is no file to write at the start. Capture `started_at` in memory before running (`date "+%Y-%m-%d %H:%M:%S%:z"`); when writing the Activity Log at the end, **append both lines at once** — the started line uses `started_at`, the done line uses the completion time, both sharing the base action (started line action gets a ` [started]` suffix, note `started`):
+
+```
+- {started_at} — **Import Dependabot [started]** by {agent} — started
+- {done_at} — **Import Dependabot** by {agent} — {completion summary}
+```
+
+`ai task log` pairs the two by base action onto one row (in progress → done). See the "Activity Log started / done dual-marker convention" in `.agents/rules/task-management.md`.
+
 ## Execution Flow
 
 ### 1. Retrieve Alert Information

package/templates/.agents/skills/import-dependabot/SKILL.zh-CN.md

@@ -17,6 +17,17 @@ description: "导入 Dependabot 安全告警并创建修复任务"
 
 > 如果 `{task-id}` 入参匹配 `^[#]?[0-9]+$`（裸数字或带 `#` 前缀），先读取 `.agents/rules/task-short-id.md` 的「SKILL 入参解析」段执行解析；后续命令视 `{task-id}` 为解析后的全长 `TASK-YYYYMMDD-HHMMSS` 形式。
 
+## 步骤开始：记录开始时间
+
+本技能会**创建** task.md，开始时尚无文件可写。先在内存记录开始时间 `started_at`（`date "+%Y-%m-%d %H:%M:%S%:z"`）；在最后写活动日志时**一次性补两条**——started 行用 `started_at`、done 行用完成时间，二者同基名（started 行 action 加 ` [started]` 后缀、note 用 `started`）：
+
+```
+- {started_at} — **Import Dependabot [started]** by {agent} — started
+- {done_at} — **Import Dependabot** by {agent} — {完成说明}
+```
+
+`ai task log` 会按基名把两条配对成一行（进行中 → 已完成）。约定见 `.agents/rules/task-management.md` 的「Activity Log started / done 双标记约定」。
+
 ## 执行流程
 
 ### 1. 获取告警信息

package/templates/.agents/skills/import-issue/SKILL.en.md

@@ -17,6 +17,22 @@ Import the specified Issue and create a task. Argument: issue number.
 
 > If `{task-id}` matches `^[#]?[0-9]+$` (bare numeric or `#`-prefixed), follow the "SKILL parameter resolver" section of `.agents/rules/task-short-id.md`; treat `{task-id}` as the resolved full `TASK-YYYYMMDD-HHMMSS` form for every downstream command.
 
+## Step Start: Capture the Start Time
+
+This skill **creates** task.md, so there is no file to write at the start. Capture `started_at` in memory before running (`date "+%Y-%m-%d %H:%M:%S%:z"`); when writing the Activity Log at the end, **append both lines at once** — the started line uses `started_at`, the done line uses the completion time, both sharing the base action (started line action gets a ` [started]` suffix, note `started`). The base action must match the actual import scenario:
+
+```
+# Scenario B: new Issue import
+- {started_at} — **Import Issue [started]** by {agent} — started
+- {done_at} — **Import Issue** by {agent} — {completion summary}
+
+# Scenario C: recovery from historical Issue comments
+- {started_at} — **Import Issue (Recovered) [started]** by {agent} — started
+- {done_at} — **Import Issue (Recovered)** by {agent} — {completion summary}
+```
+
+`ai task log` pairs the two by base action onto one row (in progress → done). See the "Activity Log started / done dual-marker convention" in `.agents/rules/task-management.md`.
+
 ## Execution Flow
 
 ### 1. Retrieve Issue Information

package/templates/.agents/skills/import-issue/SKILL.zh-CN.md

@@ -17,6 +17,22 @@ description: "从 Issue 导入并创建任务"
 
 > 如果 `{task-id}` 入参匹配 `^[#]?[0-9]+$`（裸数字或带 `#` 前缀），先读取 `.agents/rules/task-short-id.md` 的「SKILL 入参解析」段执行解析；后续命令视 `{task-id}` 为解析后的全长 `TASK-YYYYMMDD-HHMMSS` 形式。
 
+## 步骤开始：记录开始时间
+
+本技能会**创建** task.md，开始时尚无文件可写。先在内存记录开始时间 `started_at`（`date "+%Y-%m-%d %H:%M:%S%:z"`）；在最后写活动日志时**一次性补两条**——started 行用 `started_at`、done 行用完成时间，二者同基名（started 行 action 加 ` [started]` 后缀、note 用 `started`）。基名必须跟实际导入场景一致：
+
+```
+# 场景 B：新 Issue 导入
+- {started_at} — **Import Issue [started]** by {agent} — started
+- {done_at} — **Import Issue** by {agent} — {完成说明}
+
+# 场景 C：从历史 Issue 评论恢复
+- {started_at} — **Import Issue (Recovered) [started]** by {agent} — started
+- {done_at} — **Import Issue (Recovered)** by {agent} — {完成说明}
+```
+
+`ai task log` 会按基名把两条配对成一行（进行中 → 已完成）。约定见 `.agents/rules/task-management.md` 的「Activity Log started / done 双标记约定」。
+
 ## 执行流程
 
 ### 1. 获取 Issue 信息

package/templates/.agents/skills/plan-task/SKILL.en.md

@@ -31,6 +31,16 @@ Before the state check is complete, do not make external-state assertions such a
 
 > If `{task-id}` matches `^[#]?[0-9]+$` (bare numeric or `#`-prefixed), follow the "SKILL parameter resolver" section of `.agents/rules/task-short-id.md`; treat `{task-id}` as the resolved full `TASK-YYYYMMDD-HHMMSS` form for every downstream command.
 
+## Step Start: Write the started Marker
+
+After prerequisites pass and before this round's first artifact action, append a started marker to task.md `## Activity Log` (same base action as this round's done entry plus a ` [started]` suffix, note `started`):
+
+```
+- {YYYY-MM-DD HH:mm:ss±HH:MM} — **Plan Task (Round {N}) [started]** by {agent} — started
+```
+
+`ai task log` pairs it with the done entry written on completion (step 7) onto one row (in progress → done). Format and pairing rules: see the "Activity Log started / done dual-marker convention" in `.agents/rules/task-management.md`.
+
 ## Steps
 
 ### 1. Verify Prerequisites
@@ -66,7 +76,7 @@ Use it to understand:
 - identified technical risks
 - effort and complexity assessment
 
-**Round ≥ 2: respond to the prior review (only when a review artifact exists)**: if the task directory contains `review-plan.md` / `review-plan-r{N}.md`, read the highest-round review report; add a `## Response to Prior Review` section to this round's plan artifact, and for each finding verify it via Read/Grep before acting (holds → accept and fix; judged hallucinated/unfounded → rebut with counter-evidence rather than defaulting to compliance); record any open disagreement under `## Open Questions`. Round 1 has no review, so skip this section.
+**Round ≥ 2: respond to the prior review (only when a review artifact exists)**: if the task directory contains `review-plan.md` / `review-plan-r{N}.md`, read the highest-round review report; add a `## Response to Prior Review` section to this round's plan artifact, and for each finding verify it via Read/Grep, then dispose of it with one of the four states in `.agents/rules/review-handshake.md` (`accepted` / `adjusted` / `refuted` / `cannot-judge`) — every state needs commensurate evidence, never defaulting to compliance; write the disposition back to the matching row in the task.md disagreement ledger (stage=plan, round +1). Record any open disagreement under `## Open Questions`. Round 1 has no review, so skip this section.
 
 ### 4. Understand the Problem
 
@@ -87,6 +97,8 @@ Follow the `technical-design` step in `.agents/workflows/feature-development.yam
 - [ ] Define the verification strategy (tests, manual checks)
 - [ ] Assess impact and risks
 
+When this round introduces a new key design decision, record it under `## Open Questions` according to `.agents/rules/no-mid-flow-questions.md` and write the matching `HD-` ledger row.
+
 **Design principles**:
 1. **Architectural soundness**: choose the structurally correct approach; diff size is not the primary criterion. Do not pile changes onto an unsound structure just to keep the diff small
 2. **Simplicity**: given a sound architecture, prefer the simplest approach and avoid over-engineering

package/templates/.agents/skills/plan-task/SKILL.zh-CN.md

@@ -31,6 +31,16 @@ tail .agents/workspace/active/{task-id}/task.md
 
 > 如果 `{task-id}` 入参匹配 `^[#]?[0-9]+$`（裸数字或带 `#` 前缀），先读取 `.agents/rules/task-short-id.md` 的「SKILL 入参解析」段执行解析；后续命令视 `{task-id}` 为解析后的全长 `TASK-YYYYMMDD-HHMMSS` 形式。
 
+## 步骤开始：写入 started 标记
+
+确认前置条件后、本轮第一个产出动作之前，向 task.md `## 活动日志` 追加一条 started 标记（与本轮 done 条目同基名 + ` [started]` 后缀，note 用 `started`）：
+
+```
+- {YYYY-MM-DD HH:mm:ss±HH:MM} — **Plan Task (Round {N}) [started]** by {agent} — started
+```
+
+`ai task log` 会把它与步骤完成时（步骤 7）写入的 done 条目配对成一行（进行中 → 已完成）。格式与配对规则见 `.agents/rules/task-management.md` 的「Activity Log started / done 双标记约定」。
+
 ## 执行步骤
 ### 1. 验证前置条件
 
@@ -65,7 +75,7 @@ tail .agents/workspace/active/{task-id}/task.md
 - 已识别的技术风险
 - 工作量和复杂度评估
 
-**Round ≥ 2：响应上一轮审查（仅当存在审查产物时）**：若任务目录存在 `review-plan.md` / `review-plan-r{N}.md`，读取最高轮次的审查报告；在本轮方案产物中新增 `## 对上一轮审查的响应` 段，对每条发现先 Read/Grep 核实再处置（成立 → 接受并修正；判定为幻觉/不成立 → 附反证反驳，不默认顺从），未决分歧写入 `## 未决问题`。Round 1 无审查，跳过本段。
+**Round ≥ 2：响应上一轮审查（仅当存在审查产物时）**：若任务目录存在 `review-plan.md` / `review-plan-r{N}.md`，读取最高轮次的审查报告；在本轮方案产物中新增 `## 对上一轮审查的响应` 段，对每条发现先 Read/Grep 核实，再按 `.agents/rules/review-handshake.md` 的四态（`accepted` / `adjusted` / `refuted` / `cannot-judge`）处置——每态都要附相称证据，不默认顺从；并把处置回写 task.md `## 审查分歧账本` 对应行（stage=plan，round +1）。未决分歧写入 `## 未决问题`。Round 1 无审查，跳过本段。
 
 ### 4. 理解问题
 
@@ -86,6 +96,8 @@ tail .agents/workspace/active/{task-id}/task.md
 - [ ] 定义验证策略（测试、手动检查）
 - [ ] 评估方案的影响和风险
 
+遇到本轮新增的关键设计决策时，按 `.agents/rules/no-mid-flow-questions.md` 判据写入 `## 未决问题`，并回写 `HD-` 账本行。
+
 **设计原则**：
 1. **架构合理性**：选择结构正确的方案，改动大小不是首要依据。不要为了减少 diff 而在不合理的结构上叠加
 2. **简洁性**：在架构合理的前提下，优先选择最简方案，避免过度设计

package/templates/.agents/skills/plan-task/config/verify.en.json

@@ -36,6 +36,9 @@
       "expected_action_pattern": "(Plan Task|Technical Design) \\(Round \\d+\\)",
       "freshness_minutes": 30
     },
+    "review-ledger": {
+      "stage_scope": ["analysis"]
+    },
     "platform-sync": {
       "when": "issue_number_exists",
       "expected_status_label": "status: pending-design-work",

package/templates/.agents/skills/plan-task/config/verify.zh-CN.json

@@ -36,6 +36,9 @@
       "expected_action_pattern": "(Plan Task|Technical Design) \\(Round \\d+\\)",
       "freshness_minutes": 30
     },
+    "review-ledger": {
+      "stage_scope": ["analysis"]
+    },
     "platform-sync": {
       "when": "issue_number_exists",
       "expected_status_label": "status: pending-design-work",

package/templates/.agents/skills/restore-task/SKILL.en.md

@@ -20,6 +20,16 @@ Version stamp rule: when creating or updating `task.md` frontmatter, read `.agen
 
 > If `{task-id}` matches `^[#]?[0-9]+$` (bare numeric or `#`-prefixed), follow the "SKILL parameter resolver" section of `.agents/rules/task-short-id.md`; treat `{task-id}` as the resolved full `TASK-YYYYMMDD-HHMMSS` form for every downstream command.
 
+## Step Start: Write the started Marker
+
+After prerequisites pass and before this step's first artifact action, append a started marker to task.md `## Activity Log` (same base action as this step's done entry plus a ` [started]` suffix, note `started`):
+
+```
+- {YYYY-MM-DD HH:mm:ss±HH:MM} — **Restore Task [started]** by {agent} — started
+```
+
+`ai task log` pairs it with the done entry written on completion onto one row (in progress → done). See the "Activity Log started / done dual-marker convention" in `.agents/rules/task-management.md`.
+
 ## Steps
 
 ### 1. Verify Input and Environment

package/templates/.agents/skills/restore-task/SKILL.zh-CN.md

@@ -20,6 +20,16 @@ description: "从平台 Issue 评论还原本地任务文件"
 
 > 如果 `{task-id}` 入参匹配 `^[#]?[0-9]+$`（裸数字或带 `#` 前缀），先读取 `.agents/rules/task-short-id.md` 的「SKILL 入参解析」段执行解析；后续命令视 `{task-id}` 为解析后的全长 `TASK-YYYYMMDD-HHMMSS` 形式。
 
+## 步骤开始：写入 started 标记
+
+确认前置条件后、本步骤第一个产出动作之前，向 task.md `## 活动日志` 追加一条 started 标记（与本步骤 done 条目同基名 + ` [started]` 后缀，note 用 `started`）：
+
+```
+- {YYYY-MM-DD HH:mm:ss±HH:MM} — **Restore Task [started]** by {agent} — started
+```
+
+`ai task log` 会把它与完成时写入的 done 条目配对成一行（进行中 → 已完成）。约定见 `.agents/rules/task-management.md` 的「Activity Log started / done 双标记约定」。
+
 ## 执行步骤
 ### 1. 验证输入与环境
 

package/templates/.agents/skills/review-analysis/SKILL.en.md

@@ -30,6 +30,16 @@ tail .agents/workspace/active/{task-id}/task.md
 
 > If `{task-id}` matches `^[#]?[0-9]+$` (bare numeric or `#`-prefixed), follow the "SKILL parameter resolver" section of `.agents/rules/task-short-id.md`; treat `{task-id}` as the resolved full `TASK-YYYYMMDD-HHMMSS` form for every downstream command.
 
+## Step Start: Write the started Marker
+
+After prerequisites pass and before this round's first artifact action, append a started marker to task.md `## Activity Log` (same base action as this round's done entry plus a ` [started]` suffix, note `started`):
+
+```
+- {YYYY-MM-DD HH:mm:ss±HH:MM} — **Review Analysis (Round {N}) [started]** by {agent} — started
+```
+
+`ai task log` pairs it with the done entry written when the review completes onto one row (in progress → done). Format and pairing rules: see the "Activity Log started / done dual-marker convention" in `.agents/rules/task-management.md`.
+
 ## Steps
 
 ### 1. Verify Prerequisites

package/templates/.agents/skills/review-analysis/SKILL.zh-CN.md

@@ -32,6 +32,16 @@ tail .agents/workspace/active/{task-id}/task.md
 
 > 如果 `{task-id}` 入参匹配 `^[#]?[0-9]+$`（裸数字或带 `#` 前缀），先读取 `.agents/rules/task-short-id.md` 的「SKILL 入参解析」段执行解析；后续命令视 `{task-id}` 为解析后的全长 `TASK-YYYYMMDD-HHMMSS` 形式。
 
+## 步骤开始：写入 started 标记
+
+确认前置条件后、本轮第一个产出动作之前，向 task.md `## 活动日志` 追加一条 started 标记（与本轮 done 条目同基名 + ` [started]` 后缀，note 用 `started`）：
+
+```
+- {YYYY-MM-DD HH:mm:ss±HH:MM} — **Review Analysis (Round {N}) [started]** by {agent} — started
+```
+
+`ai task log` 会把它与审查完成时写入的 done 条目配对成一行（进行中 → 已完成）。格式与配对规则见 `.agents/rules/task-management.md` 的「Activity Log started / done 双标记约定」。
+
 ## 执行步骤
 ### 1. 验证前置条件
 

package/templates/.agents/skills/review-analysis/config/verify.en.json

@@ -24,7 +24,8 @@
         "Conclusion and Recommendations",
         "State Check",
         "Evidence",
-        "Self-Doubt"
+        "Self-Doubt",
+        "Review Disagreement Ledger Writeback"
       ],
       "required_patterns": [
         "^### Approval Decision$",

package/templates/.agents/skills/review-analysis/config/verify.zh-CN.json

@@ -24,7 +24,8 @@
         "结论与建议",
         "状态核对",
         "证据原文",
-        "自我质疑"
+        "自我质疑",
+        "审查分歧账本回写"
       ],
       "required_patterns": [
         "^### 审查决定$",

package/templates/.agents/skills/review-analysis/reference/output-templates.en.md

@@ -16,12 +16,13 @@ Rules:
 - If `Blocker > 0`, never use an approved template
 - Never count env-blocked items as blocker / major / minor or use them to trigger Scenario B/C/D
 - The selected scenario must include all TUI command formats
+- The count line always shows 4 numbers: the first three (Blockers / Major / Minor) must be 0 to proceed; the fourth, `Human-decision` (`{h}`), is the number of rows in task.md `## 审查分歧账本` with `stage=analysis` and `status=needs-human-decision` — a "pending human ruling" item that need not be zero and does not participate in scenario selection
 
 ### Scenario A: Approved with no findings
 
 ```text
 Task {task-id} requirement analysis review completed. Verdict: approved.
-- Blockers: 0 | Major issues: 0 | Minor issues: 0[ | Environment-blocked: {n} (outside AI fix scope)]
+- Blockers: 0 | Major issues: 0 | Minor issues: 0 | Human-decision: {h}
 [- Review report: .agents/workspace/active/{task-id}/{review-artifact}]
 
 Next step - write the technical plan:
@@ -37,7 +38,7 @@ Reminder: env-blocked items belong in the PR description manual verification che
 
 ```text
 Task {task-id} requirement analysis review completed. Verdict: approved.
-- Blockers: 0 | Major issues: {n} | Minor issues: {n}[ | Environment-blocked: {n} (outside AI fix scope)]
+- Blockers: 0 | Major issues: {n} | Minor issues: {n} | Human-decision: {h}
 - Review report: .agents/workspace/active/{task-id}/{review-artifact}
 
 Next step - revise analysis before continuing (recommended):
@@ -58,7 +59,7 @@ Reminder: env-blocked items belong in the PR description manual verification che
 
 ```text
 Task {task-id} requirement analysis review completed. Verdict: changes requested.
-- Blockers: {n} | Major issues: {n} | Minor issues: {n}[ | Environment-blocked: {n} (outside AI fix scope)]
+- Blockers: {n} | Major issues: {n} | Minor issues: {n} | Human-decision: {h}
 - Review report: .agents/workspace/active/{task-id}/{review-artifact}
 
 Next step - revise requirement analysis:
@@ -74,7 +75,7 @@ Reminder: env-blocked items belong in the PR description manual verification che
 
 ```text
 Task {task-id} requirement analysis review completed. Verdict: rejected, fresh analysis or requirement clarification required.
-- Blockers: {n} | Major issues: {n} | Minor issues: {n}[ | Environment-blocked: {n} (outside AI fix scope)]
+- Blockers: {n} | Major issues: {n} | Minor issues: {n} | Human-decision: {h}
 - Review report: .agents/workspace/active/{task-id}/{review-artifact}
 
 Next step - re-analyze:

package/templates/.agents/skills/review-analysis/reference/output-templates.zh-CN.md

@@ -16,12 +16,13 @@
 - 只要 `Blocker > 0`，就绝对不能输出通过模板
 - env-blocked 项绝对不能被计入 blocker / major / minor 计数，也不能用作触发场景 B/C/D 的依据
 - 所选场景中必须包含所有 TUI 命令格式
+- 计数行固定显示 4 个数字：前三项（阻塞 / 主要 / 次要）必须为 0 才进下一步；第四项 `人工裁决`（`{h}`）= task.md `## 审查分歧账本` 中 `stage=analysis` 且 `status=needs-human-decision` 的行数，是「待人裁」项、不要求归零，也不参与场景判断
 
 ### 场景 A：通过且无问题
 
 ```text
 任务 {task-id} 需求分析审查完成。结论：通过。
-- 阻塞项：0 | 主要问题：0 | 次要问题：0[ | 环境性遗留：{n}（不在 AI 修复范围）]
+- 阻塞项：0 | 主要问题：0 | 次要问题：0 | 人工裁决：{h}
 [- 审查报告：.agents/workspace/active/{task-id}/{review-artifact}]
 
 下一步 - 编写技术方案：
@@ -37,7 +38,7 @@
 
 ```text
 任务 {task-id} 需求分析审查完成。结论：通过。
-- 阻塞项：0 | 主要问题：{n} | 次要问题：{n}[ | 环境性遗留：{n}（不在 AI 修复范围）]
+- 阻塞项：0 | 主要问题：{n} | 次要问题：{n} | 人工裁决：{h}
 - 审查报告：.agents/workspace/active/{task-id}/{review-artifact}
 
 下一步 - 修订分析后继续（推荐）：
@@ -58,7 +59,7 @@
 
 ```text
 任务 {task-id} 需求分析审查完成。结论：需要修改。
-- 阻塞项：{n} | 主要问题：{n} | 次要问题：{n}[ | 环境性遗留：{n}（不在 AI 修复范围）]
+- 阻塞项：{n} | 主要问题：{n} | 次要问题：{n} | 人工裁决：{h}
 - 审查报告：.agents/workspace/active/{task-id}/{review-artifact}
 
 下一步 - 修订需求分析：
@@ -74,7 +75,7 @@
 
 ```text
 任务 {task-id} 需求分析审查完成。结论：拒绝，需要重新分析或补充需求澄清。
-- 阻塞项：{n} | 主要问题：{n} | 次要问题：{n}[ | 环境性遗留：{n}（不在 AI 修复范围）]
+- 阻塞项：{n} | 主要问题：{n} | 次要问题：{n} | 人工裁决：{h}
 - 审查报告：.agents/workspace/active/{task-id}/{review-artifact}
 
 下一步 - 重新分析：

package/templates/.agents/skills/review-analysis/reference/report-template.en.md

@@ -59,6 +59,10 @@ Use this template when writing `review-analysis.md` or `review-analysis-r{N}.md`
 > If this round has no env-blocked findings, keep the section heading and write "None".
 
 
+## Review Disagreement Ledger Writeback
+
+> Upsert each finding this round into the task.md disagreement ledger: append an `open` row for new findings (id prefix `AN-`, stage=analysis); per the hand-back duty set the executor's prior-round responses to `confirmed` / back to `open` / `needs-human-decision`. State machine and evidence rules: `.agents/rules/review-handshake.md`.
+
 ## Evidence
 
 > Pair each "I verified X" claim with the corresponding raw tool output; the gate only checks that this section exists and at least one `$ ` line is present. Every Blocker must be backed by a reproducible command (rg/grep/sed/nl) and its raw output; a judgment that cannot be reproduced must be downgraded or moved to Self-Doubt.

package/templates/.agents/skills/review-analysis/reference/report-template.zh-CN.md

@@ -60,6 +60,10 @@
 > 如本轮无 env-blocked 项，保留段落标题并写「（无）」。
 
 
+## 审查分歧账本回写
+
+> 把本轮每条 finding upsert 到 task.md `## 审查分歧账本`：新 finding 追加 `open` 行（id 前缀 `AN-`，stage=analysis），对执行方上一轮响应按回交义务改 `confirmed` / 置回 `open` / `needs-human-decision`。状态机与证据规则见 `.agents/rules/review-handshake.md`。
+
 ## 证据原文
 
 > 每条“我验证了 X”断言都要配对对应 tool output 原文；gate 仅校验本段存在和至少一行 `$ `。每条 Blocker 必须配可复现命令（rg/grep/sed/nl）及其原文；无法复现的判断须降级或移入「自我质疑」。

package/templates/.agents/skills/review-analysis/reference/review-criteria.en.md

@@ -13,6 +13,7 @@ Follow the `analysis-review` step in `.agents/workflows/feature-development.yaml
 - [ ] Risks, edge cases, and open questions are recorded
 - [ ] The design stage has enough input to proceed
 - [ ] The analysis matches the original Issue or user request
+- [ ] The reviewer checked whether the executor missed any key design decision that should be upgraded to `[needs-human-decision]`
 - [ ] Every blocker is backed by reproducible grep/sed/nl evidence; conclusions not directly verified are declared under Self-Doubt
 
 **Common anti-examples**:

package/templates/.agents/skills/review-analysis/reference/review-criteria.zh-CN.md

@@ -13,6 +13,7 @@
 - [ ] 风险、边界情况和开放问题是否记录
 - [ ] 后续设计阶段是否有足够输入
 - [ ] 与原始 Issue / 用户需求是否一致
+- [ ] 已复核执行方是否漏标应升级为 `[needs-human-decision]` 的关键设计决策
 - [ ] 每条 blocker 都配可复现的 grep/sed/nl 证据，未直接验证的结论已在「自我质疑」声明
 
 **常见反例**：

package/templates/.agents/skills/review-code/SKILL.en.md

@@ -41,6 +41,16 @@ Before the state check is complete, do not make external-state assertions such a
 
 > If `{task-id}` matches `^[#]?[0-9]+$` (bare numeric or `#`-prefixed), follow the "SKILL parameter resolver" section of `.agents/rules/task-short-id.md`; treat `{task-id}` as the resolved full `TASK-YYYYMMDD-HHMMSS` form for every downstream command.
 
+## Step Start: Write the started Marker
+
+After prerequisites pass and before this round's first artifact action, append a started marker to task.md `## Activity Log` (same base action as this round's done entry plus a ` [started]` suffix, note `started`):
+
+```
+- {YYYY-MM-DD HH:mm:ss±HH:MM} — **Review Code (Round {N}) [started]** by {agent} — started
+```
+
+`ai task log` pairs it with the done entry written when the review completes onto one row (in progress → done). Format and pairing rules: see the "Activity Log started / done dual-marker convention" in `.agents/rules/task-management.md`.
+
 ## Steps
 
 ### 1. Verify Prerequisites
@@ -61,7 +71,10 @@ Read the highest-round code artifact and, if present, the highest-round fix arti
 
 ### 4. Perform the Review
 
-Follow `.agents/workflows/feature-development.yaml` and inspect `git diff` for the full change context.
+Follow `.agents/workflows/feature-development.yaml` and inspect the full change context:
+- `git diff --binary HEAD -- <post-review-globs>` for tracked changes
+- `git ls-files -o --exclude-standard -z -- <post-review-globs>` for untracked new files
+- `node .agents/scripts/review-diff-fingerprint.js worktree HEAD` for the reviewed diff fingerprint; write it into the report
 
 > Detailed review criteria, severity rules, and reviewer expectations live in `reference/review-criteria.md`. Read `reference/review-criteria.md` before reviewing.
 > Test review gate: when `git diff` touches test files, read `.agents/rules/testing-discipline.md` first and check it item by item, especially "do not add negative assertions when a positive assertion already covers the behavior".

package/templates/.agents/skills/review-code/SKILL.zh-CN.md

@@ -41,6 +41,16 @@ tail .agents/workspace/active/{task-id}/task.md
 
 > 如果 `{task-id}` 入参匹配 `^[#]?[0-9]+$`（裸数字或带 `#` 前缀），先读取 `.agents/rules/task-short-id.md` 的「SKILL 入参解析」段执行解析；后续命令视 `{task-id}` 为解析后的全长 `TASK-YYYYMMDD-HHMMSS` 形式。
 
+## 步骤开始：写入 started 标记
+
+确认前置条件后、本轮第一个产出动作之前，向 task.md `## 活动日志` 追加一条 started 标记（与本轮 done 条目同基名 + ` [started]` 后缀，note 用 `started`）：
+
+```
+- {YYYY-MM-DD HH:mm:ss±HH:MM} — **Review Code (Round {N}) [started]** by {agent} — started
+```
+
+`ai task log` 会把它与审查完成时写入的 done 条目配对成一行（进行中 → 已完成）。格式与配对规则见 `.agents/rules/task-management.md` 的「Activity Log started / done 双标记约定」。
+
 ## 执行步骤
 ### 1. 验证前置条件
 
@@ -60,7 +70,10 @@ tail .agents/workspace/active/{task-id}/task.md
 
 ### 4. 执行审查
 
-遵循 `.agents/workflows/feature-development.yaml`，并同时检查 `git diff` 获取完整变更上下文。
+遵循 `.agents/workflows/feature-development.yaml`，并同时检查完整变更上下文：
+- `git diff --binary HEAD -- <post-review-globs>` 覆盖已跟踪变更
+- `git ls-files -o --exclude-standard -z -- <post-review-globs>` 覆盖未跟踪新文件
+- `node .agents/scripts/review-diff-fingerprint.js worktree HEAD` 生成审查差异指纹，并写入报告
 
 > 详细审查标准、严重程度划分和 reviewer 关注点见 `reference/review-criteria.md`。执行此步骤前先读取 `reference/review-criteria.md`。
 > 测试审查硬门禁：当 `git diff` 触及测试文件时，必须先读取 `.agents/rules/testing-discipline.md` 并逐条核对（尤其"正向已覆盖时不应再加反向断言"）。

package/templates/.agents/skills/review-code/config/verify.en.json

@@ -24,12 +24,15 @@
         "Conclusion and Recommendations",
         "State Check",
         "Evidence",
-        "Self-Doubt"
+        "Self-Doubt",
+        "Review Disagreement Ledger Writeback"
       ],
       "required_patterns": [
         "^### Approval Decision$",
         "^\\$ ",
-        "^- \\*\\*Overall Verdict\\*\\*[:：]\\s*(?:通过|需要修改|拒绝|Approved|Changes Requested|Rejected)\\s*$"
+        "^- \\*\\*Overall Verdict\\*\\*[:：]\\s*(?:通过|需要修改|拒绝|Approved|Changes Requested|Rejected)\\s*$",
+        "^- \\*\\*Review Baseline Commit\\*\\*[:：]\\s*\\S",
+        "^- \\*\\*Reviewed Diff Fingerprint\\*\\*[:：]\\s*sha256:[0-9a-f]{64}\\s*$"
       ],
       "freshness_minutes": 30
     },