@fitlab-ai/agent-infra 0.9.2 → 0.9.4

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

@@ -7,6 +7,9 @@ description: >
 ---
 
 # Technical Plan Review
+> `--agent` values follow the "Collaborator Token Specification" in `.agents/rules/task-management.md`: standard AI short tokens (`claude`/`codex`/`antigravity`/`opencode`/`cursor`), long-name normalization (`claude-code`->`claude`, `antigravity-cli`->`antigravity`), or the `human` manual exception.
+
+If the entry operands contain `--orchestrated`, bind `{execution-flag}` to `--orchestrated` and forward it unchanged to both the summary finalizer and completed event; otherwise bind it to an empty value. Never infer it from `orchestration.json`, environment variables, or prior artifacts.
 
 Review the latest plan artifact and produce `review-plan.md` or `review-plan-r{N}.md`.
 
@@ -35,7 +38,7 @@ agent-infra-internal task-snapshot {task-id} --format text
 
 ## Step Start: Write the started Marker
 
-After resolving the artifact context and before this round's first artifact action, run `agent-infra-internal task-event {task-id} review-plan.started --agent {agent}`.
+After resolving the artifact context and before this round's first artifact action, run `agent-infra-internal task-event {task-id} review-plan.started --agent {standard-agent-token}`.
 
 ## Steps
 
@@ -66,7 +69,7 @@ Create `.agents/workspace/active/{task-id}/{review-artifact}`.
 
 ### 6. Update Task Status
 
-After the report, submit each new finding with `agent-infra-internal task-ledger {task-id} finding-upsert --stage plan --review-artifact {review-artifact} --ordinal {n} --severity {blocker|major|minor} --evidence {review-artifact}#{anchor}`; submit prior-response dispositions with `finding-review --id {ledger-id} --status {confirmed|closed|open|needs-human-decision} --evidence {evidence}`. Do not scan ids or edit ledger rows. After all ledger writes, call `agent-infra-internal task-review {task-id} finalize-summary --stage plan --artifact {review-artifact}` exactly once.
+After the report, submit each new finding with `agent-infra-internal task-ledger {task-id} finding-upsert --stage plan --review-artifact {review-artifact} --ordinal {n} --severity {blocker|major|minor} --evidence {review-artifact}#{anchor}`; submit prior-response dispositions with `finding-review --id {ledger-id} --status {confirmed|closed|open|needs-human-decision} --evidence {evidence}`. Do not scan ids or edit ledger rows. After all ledger writes, call `agent-infra-internal task-review {task-id} finalize-summary --stage plan --artifact {review-artifact} {execution-flag}` exactly once.
 
 Bind and reuse this structured mapping from that one response:
 
@@ -76,11 +79,11 @@ Bind and reuse this structured mapping from that one response:
 {unresolved-minor} = stageStatus.unresolvedFindingCounts.minor
 ```
 
-The intent atomically finalizes the report summary and returns the same ledger snapshot. Do not call `stage-status`, replace placeholders manually, or rescan the finding list. If finalization fails or a response field is missing, stop before the completion event. Derive the verdict and next-step branch from the same response's `stageStatus.canAdvance`: only `canAdvance=true` permits `approved`; otherwise use `changes-requested` or `rejected`. Then run `agent-infra-internal task-event {task-id} review-plan.completed --agent {agent} --artifact {review-artifact} --verdict {approved|changes-requested|rejected} --blockers {unresolved-blockers} --major {unresolved-major} --minor {unresolved-minor} --manual-validation {n}`.
+The intent atomically finalizes the report summary and returns the same ledger snapshot. Do not call `stage-status`, replace placeholders manually, or rescan the finding list. If finalization fails or a response field is missing, stop before the completion event. Derive the verdict and next-step branch from the same response's `stageStatus.canAdvance`: only `canAdvance=true` permits `approved`; otherwise use `changes-requested` or `rejected`. Then run `agent-infra-internal task-event {task-id} review-plan.completed --agent {standard-agent-token} --artifact {review-artifact} --verdict {approved|changes-requested|rejected} --blockers {unresolved-blockers} --major {unresolved-major} --minor {unresolved-minor} --manual-validation {n} {execution-flag}`.
 
 `manual-validation` is the data source for the `Manual-validation` count folded into review rows in `ai task log`; do not add a parallel manual-verification field.
 
-If task.md has a valid `issue_number`, run `agent-infra-internal platform-comment sync {task-id} --kind task --agent {agent}`, then `agent-infra-internal platform-comment sync {task-id} --kind artifact --artifact {review-artifact} --agent {agent}`.
+If task.md has a valid `issue_number`, run `agent-infra-internal platform-comment sync {task-id} --kind task --agent {standard-agent-token}`, then `agent-infra-internal platform-comment sync {task-id} --kind artifact --artifact {review-artifact} --agent {standard-agent-token}`.
 
 ### 7. Run Completion Gate
 

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

@@ -7,6 +7,9 @@ description: >
 ---
 
 # 技术方案审查
+> `--agent` 取值见 `.agents/rules/task-management.md`「合作者 token 规范」：标准 AI 短名（`claude`/`codex`/`antigravity`/`opencode`/`cursor`）、长名归一化（`claude-code`→`claude`、`antigravity-cli`→`antigravity`）或人工例外 `human`。
+
+若入口业务操作数包含 `--orchestrated`，绑定 `{execution-flag}` = `--orchestrated` 并原样转发给 summary finalizer 与 completed 事件；否则绑定为空。不得从 `orchestration.json`、环境变量或历史产物推断该标记。
 
 审查最新技术方案产物，并产出 `review-plan.md` 或 `review-plan-r{N}.md`。
 
@@ -37,7 +40,7 @@ agent-infra-internal task-snapshot {task-id} --format text
 
 ## 步骤开始：声明 started 事件
 
-确认前置条件和产物上下文后、本轮第一个产出动作之前执行 `agent-infra-internal task-event {task-id} review-plan.started --agent {agent}`。
+确认前置条件和产物上下文后、本轮第一个产出动作之前执行 `agent-infra-internal task-event {task-id} review-plan.started --agent {standard-agent-token}`。
 
 ## 执行步骤
 ### 1. 验证前置条件
@@ -69,7 +72,7 @@ agent-infra-internal task-snapshot {task-id} --format text
 
 ### 6. 更新任务状态
 
-报告完成后，新 finding 逐条调用 `agent-infra-internal task-ledger {task-id} finding-upsert --stage plan --review-artifact {review-artifact} --ordinal {n} --severity {blocker|major|minor} --evidence {review-artifact}#{anchor}`；复核上一轮响应时调用 `finding-review --id {ledger-id} --status {confirmed|closed|open|needs-human-decision} --evidence {相称证据}`。不得扫描编号或手写账本行。全部账本写入完成后只调用一次 `agent-infra-internal task-review {task-id} finalize-summary --stage plan --artifact {review-artifact}`。
+报告完成后，新 finding 逐条调用 `agent-infra-internal task-ledger {task-id} finding-upsert --stage plan --review-artifact {review-artifact} --ordinal {n} --severity {blocker|major|minor} --evidence {review-artifact}#{anchor}`；复核上一轮响应时调用 `finding-review --id {ledger-id} --status {confirmed|closed|open|needs-human-decision} --evidence {相称证据}`。不得扫描编号或手写账本行。全部账本写入完成后只调用一次 `agent-infra-internal task-review {task-id} finalize-summary --stage plan --artifact {review-artifact} {execution-flag}`。
 
 从该次返回值绑定并复用以下结构化映射：
 
@@ -79,11 +82,11 @@ agent-infra-internal task-snapshot {task-id} --format text
 {unresolved-minor} = stageStatus.unresolvedFindingCounts.minor
 ```
 
-该 intent 原子最终化报告摘要并返回同一次账本快照；不得再调用 `stage-status`、手工替换占位符或扫描问题清单。最终化失败或返回字段缺失时，停止在完成事件之前。以同一次返回的 `stageStatus.canAdvance` 决定 verdict 和下一步：仅 `canAdvance=true` 可用 `approved`，否则必须用 `changes-requested` 或 `rejected`。随后执行 `agent-infra-internal task-event {task-id} review-plan.completed --agent {agent} --artifact {review-artifact} --verdict {approved|changes-requested|rejected} --blockers {unresolved-blockers} --major {unresolved-major} --minor {unresolved-minor} --manual-validation {n}`。
+该 intent 原子最终化报告摘要并返回同一次账本快照；不得再调用 `stage-status`、手工替换占位符或扫描问题清单。最终化失败或返回字段缺失时，停止在完成事件之前。以同一次返回的 `stageStatus.canAdvance` 决定 verdict 和下一步：仅 `canAdvance=true` 可用 `approved`，否则必须用 `changes-requested` 或 `rejected`。随后执行 `agent-infra-internal task-event {task-id} review-plan.completed --agent {standard-agent-token} --artifact {review-artifact} --verdict {approved|changes-requested|rejected} --blockers {unresolved-blockers} --major {unresolved-major} --minor {unresolved-minor} --manual-validation {n} {execution-flag}`。
 
 `manual-validation` 是 `ai task log` 中 review 行「人工校验点」（EN `Manual-validation`）计数的数据源；不要新增并行人工验证字段。
 
-如果 task.md 中存在有效的 `issue_number`，调用 `agent-infra-internal platform-comment sync {task-id} --kind task --agent {agent}`，再调用 `agent-infra-internal platform-comment sync {task-id} --kind artifact --artifact {review-artifact} --agent {agent}`；失败按 `.agents/rules/issue-sync.md` 记录 warning。
+如果 task.md 中存在有效的 `issue_number`，调用 `agent-infra-internal platform-comment sync {task-id} --kind task --agent {standard-agent-token}`，再调用 `agent-infra-internal platform-comment sync {task-id} --kind artifact --artifact {review-artifact} --agent {standard-agent-token}`；失败按 `.agents/rules/issue-sync.md` 记录 warning。
 
 ### 7. 完成校验
 

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

@@ -0,0 +1,186 @@
+---
+name: review-pr
+description: >
+  Review a pull request by evidence grade and publish a formal PR review.
+  Use when a PR needs review (regardless of whether trusted lifecycle records exist).
+  Only invoke this skill automatically when the conversation includes a resolvable PR number or PR URL.
+---
+
+# Review Pull Request
+
+Based on the target PR's lifecycle-evidence completeness and change risk, select `verify` (lightweight recheck), `audit` (evidence audit), or `reconstruct` (reconstruction review), publish the actionable review conclusion as a formal PR Review on the target PR, and retain traceable review-process evidence (`pr-review.md` / `pr-review-r{N}.md`).
+
+## Boundary / Critical Rules
+
+- Strictly follow the fixed evidence-grading decision flow: host resolution → evidence classification → freshness/alignment → risk grading → mode selection
+- `pr-review*` is an independent artifact family: it does not join the analysis/plan/code stage chain and never changes `current_step`
+- Only a formal PR Review (head SHA, conclusion, findings, receipt, Issue artifact link) is published on the PR; no second full process copy goes to a regular PR comment
+- The full process copy and the reversible `task.md` mirror are synced only to the linked Issue; `restore-task` keeps its Issue-only contract
+- Never backfill or fabricate `analysis`/`plan`/`code` lifecycle history for an external PR
+- The head SHA is recorded only in `pr-review*` and the formal PR Review; on head drift you must start a new round and never reuse an old conclusion
+- Review depth is driven by evidence completeness and change risk, never by the submitter's identity
+- After executing this skill you **must** immediately update task.md (via the `task-activity` typed intent)
+- Never run `git add` or `git commit` automatically
+- The one-shot path (no task/Issue) uses `.agents/workspace/reviews/{pr-number}/` with `recoverable: false` and carries no recovery promise
+
+## Common Rationalizations and Rebuttals
+
+| Rationalization | Rebuttal |
+|-----------------|----------|
+| "Just read the diff; no reconstruction needed" | Reading only the diff misses requirement boundaries, architecture choices, and migration strategy; `reconstruct` must first produce the minimum-sufficient reconstruction record. |
+| "The PR has no linked task; just post a report" | Block by default and require linking an Issue/task first; only an explicit "one-shot review" takes the non-recoverable fallback path. |
+| "Also post the full reconstruction to a regular PR comment for contributors" | The process copy is synced only to the Issue artifact comment; the PR keeps only the formal Review to avoid duplicate remote copies and restore-source confusion. |
+| "Commit while I'm at it" | This skill never runs `git add`/`git commit`; committing is a separate, explicitly user-triggered step. |
+
+## Step 0: State Check (pre-execution hard gate)
+
+Before making any task-state judgment or user-visible conclusion, run the state check:
+
+```bash
+agent-infra-internal task-snapshot {task-id} --format text
+```
+
+The task-anchored path records the current task directory and existing `pr-review*` rounds before starting; the one-shot path (no task) records the PR number and target directory. Paste the state-check output into the `## State Check` section of this round's `pr-review-rN.md`.
+
+## Steps
+
+### 1. Resolve the Input and the Host
+
+Resolve the target PR number `{pr-number}` (`--pr <number>`, a PR URL, or by reverse lookup from the current branch). Run:
+
+```bash
+agent-infra-internal platform-pr-review inspect --pr {pr-number} [--cwd <path>]
+agent-infra-internal pr-review-grade resolve-host --pr {pr-number} [--cwd <path>]
+```
+
+`resolve-host` returns a typed `HostResolution` (`unique` / `ambiguous` / `none`). Branch as follows:
+
+- **Unique host**: bind `{task-id}`, call `task-activity pr-review-inspect` to obtain the canonical round, prepared/open state, and latest successful review identity, then start this round using the recovery rules below (go to step 2).
+- **Ambiguous hosts**: `resolve-host` returns `ambiguous` and `decide` refuses to classify (fail closed). Stop and ask a human to pin the unique host; do not enter evidence classification.
+- **No host**: block by default and require linking an Issue/task first (show linking guidance); never auto-create/import an Issue. Only when the user explicitly chooses a "one-shot review" do process files land in `.agents/workspace/reviews/{pr-number}/` (`recoverable: false`).
+
+After the task-anchored path resolves the host and remote head, run:
+
+```bash
+agent-infra-internal task-activity {task-id} pr-review-inspect
+```
+
+- `prepared`: reuse that artifact; `open` with an unchanged head: replay start for the same artifact/head (a no-op) and resume the round.
+- `open` with a changed head: first set the old artifact Formal Review Status to `superseded`, then terminate it using the old artifact/head; continue with the next round returned by inspect.
+- No prepared/open round: create the next artifact skeleton from `reference/report-template.md`, initially recording the state check, identity, reviewed head, and `Formal Review Status: pending`.
+
+After this branch and before step 2 evidence grading, write started:
+
+```bash
+agent-infra-internal task-activity {task-id} pr-review-start --agent {agent} \
+  --artifact {pr-review-artifact} --head {head-sha}
+```
+
+The one-shot path never calls `task-activity`. For a controlled failure after started where the formal Review is known not to have been published, first set the artifact status to `aborted`, then call `pr-review-terminate --outcome aborted --reason <single-line>` with the same artifact/head. If publication outcome is uncertain, leave the round open rather than guessing aborted.
+
+### 2. Single decide: evidence enumeration + classification + risk + mode
+
+Write the host, artifact presence, head state, and the six pure-evidence risk factors to an input JSON. Use only the highest `applied` / `no-op` artifact returned by inspect as the prior review; pending/aborted/superseded/failed artifacts count only toward round continuity. Then call once:
+
+```bash
+agent-infra-internal pr-review-grade decide --input-file {decide-input.json} [--cwd <path>]
+```
+
+Returns the full `DecisionRecord` (`scenario` / `freshness` / `alignment` / `risk` / `mode` / `firstReview` / `reason`) in one call. Record the decision input and output verbatim in the `pr-review-rN.md` "Evidence List" section (AC3 traceability). When `host.kind === 'ambiguous'`, the command refuses; return to step 1's blocking exit.
+
+### 3. Generate `pr-review-rN.md`
+
+Complete this round's artifact per `reference/report-template.md`; the task-anchored path preserves the identity/head/pending status written in step 1 and does not reallocate the round. In `reconstruct` mode (or `audit` with insufficient evidence), write a "Reconstruction Context" section (requirement boundary / architecture choices / impact surface / validation coverage, see `reference/evidence-grading.md`) before the line-level findings. Record `recoverable: true|false` in the frontmatter or header (the one-shot path sets `false`). After finalizing the findings list, freeze `{verdict, blockers, major, minor}` once; the formal Review body and step 6 complete payload must consume that same object directly, never recounting report prose.
+
+### 4. Sync the Issue artifact comments (task-anchored path)
+
+Sync the task comment, then the artifact comment, and capture a stable comment URL:
+
+```bash
+agent-infra-internal platform-comment sync {task-id} --kind task --agent {agent}
+agent-infra-internal platform-comment sync {task-id} --kind artifact --artifact {pr-review-artifact} --agent {agent}
+```
+
+The one-shot path (no task/Issue) skips this step.
+
+### 5. Recheck the head and publish the formal Review
+
+Read the remote head again to confirm it did not drift during the review:
+
+```bash
+agent-infra-internal platform-pr-review inspect --pr {pr-number} [--cwd <path>]
+```
+
+If the head matches step 1, assemble the body (head SHA / conclusion / findings / receipt / Issue artifact link, **without a marker**) and publish the formal Review:
+
+```bash
+agent-infra-internal platform-pr-review publish --pr {pr-number} --scope {taskId|pr{pr-number}} --round {round} \
+  --commit {head-sha} --event {COMMENT|APPROVE|REQUEST_CHANGES} --body-file {review-body.md} [--dry-run] [--cwd <path>]
+```
+
+`publish` generates and validates the marker (first line) in core and is idempotent per marker + commit (replay is a no-op; marker hit on a different commit fails stably). On head drift, first set the old artifact status to `superseded`, then close the old round:
+
+```bash
+agent-infra-internal task-activity {task-id} pr-review-terminate --agent {agent} \
+  --artifact {pr-review-artifact} --head {head-sha} \
+  --outcome superseded --reason "head changed before publish"
+```
+
+After closure, re-run step 1 for the next canonical round; never start the new round first. If the remote outcome of `publish` is uncertain, keep the round open and recover through marker + commit idempotency on retry.
+
+### 6. Write back the publication result and task state
+
+- **Task-anchored path**: after publish returns `applied` / `no-op`, write the Review ID/URL and matching Formal Review Status into the `pr-review-rN.md` "Publication Result" section, then close the Activity Log with the exact verdict/counts frozen in step 3:
+
+  ```bash
+  agent-infra-internal task-activity {task-id} pr-review-complete --agent {agent} \
+    --artifact {pr-review-artifact} --head {head-sha} --verdict {approved|changes-requested|commented} \
+    --blockers {blockers} --major {major} --minor {minor}
+  ```
+
+  `task-activity` generates `Verdict: <result>, blockers: N, major: N, minor: N → {pr-review-artifact}` from the typed payload and atomically writes the Activity Log, `## Review Feedback` link, and version stamp through `writeTask`. It never changes `current_step` or puts receipt/head/Review URL into the NOTE.
+
+- **One-shot path (no task)**: write the Review ID/URL only into the `pr-review-rN.md` "Publication Result" section; do not call `task-activity`.
+
+### 7. Re-sync after write-back (task-anchored path)
+
+Step 6 rewrote the local `pr-review-rN.md` "Publication Result" section and task.md (Activity Log), while step 4 synced the older snapshot. `verify_comment_content` / `verify_task_comment_content` compare full content, so you must re-sync first to align local and remote. Run in order:
+
+```bash
+agent-infra-internal platform-comment sync {task-id} --kind task --agent {agent}
+agent-infra-internal platform-comment sync {task-id} --kind artifact --artifact {pr-review-artifact} --agent {agent}
+```
+
+`platform-comment sync` updates an existing marker comment in place (idempotent), so the comment URL is unchanged. The one-shot path has no task and skips this step.
+
+### 8. Run the completion verification
+
+- **Task-anchored path**:
+
+  ```bash
+  agent-infra-internal task-verify {task-id} review-pr.completed --artifact {pr-review-artifact} --format text
+  ```
+
+- **One-shot path**:
+
+  ```bash
+  agent-infra-internal pr-review-grade verify-artifact --artifact-file {pr-review-artifact} [--cwd <path>]
+  ```
+
+Exit code 0 passes; 1 means fix per the output and re-run; 2 means stop and ask for human intervention.
+
+### 9. Inform the user
+
+Pick the single exit branch per `reference/output-guidance.md` (published / blocked-requires-link / one-shot), and read `.agents/rules/next-step-output.md` before rendering the next step.
+
+## Completion Checklist
+
+- [ ] Completed the evidence-graded PR review and produced `pr-review-rN.md`
+- [ ] Published the formal Review on the target PR, bound to the reviewed head SHA
+- [ ] Synced and re-synced the Issue artifact/task comments on the task-anchored path
+- [ ] Ran the completion verification (`task-verify` or `verify-artifact`)
+- [ ] Updated task.md and appended the Activity Log entry (task-anchored path)
+
+## Stop
+
+Stop as soon as the checklist is complete. Do not commit automatically.

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

@@ -0,0 +1,186 @@
+---
+name: review-pr
+description: >
+  按证据分级审查 PR 并发布正式 Review。
+  当需要审查一个 PR（无论是否已有可信生命周期记录）时使用。
+  仅当对话包含可解析的 PR 编号或 PR URL 时才可自动调用本技能。
+---
+
+# 审查 PR
+
+根据目标 PR 的生命周期证据完整度与变更风险，选择 `verify`（轻量复核）、`audit`（证据审计）或 `reconstruct`（重建式审查），并把可执行的检视结论作为正式 PR Review 发布到目标 PR，同时保留可追溯的审查过程证据（`pr-review.md` / `pr-review-r{N}.md`）。
+
+## 行为边界 / 关键规则
+
+- 严格遵循方案中固化的证据分级主决策流：宿主解析 → 证据分类 → 新鲜度/对齐 → 风险分级 → 模式选择
+- `pr-review*` 是独立 artifact family：不接入 analysis/plan/code 阶段依赖链，不修改 `current_step`
+- PR 上只发布正式 PR Review（head SHA、结论、finding、receipt、Issue artifact 链接），不发第二份完整过程原文到 PR 普通评论
+- 完整过程原文与 `task.md` 的可逆镜像只在关联 Issue 上同步；`restore-task` 保持 Issue-only 契约
+- 不得为外部 PR 倒填或伪造已发生的 `analysis`/`plan`/`code` 生命周期历史
+- head SHA 唯一记录于 `pr-review*` 与正式 PR Review；head 漂移必须进入新轮次，不得沿用旧结论
+- 审查深度由证据完整度和变更风险决定，不由提交者身份决定
+- 执行本技能后，你**必须**立即更新 task.md（经 `task-activity` typed intent）
+- 绝不自动执行 `git add` 或 `git commit`
+- 一次性检视路径（无任务/Issue）走 `.agents/workspace/reviews/{pr-number}/`，标记 `recoverable: false`，不承担可恢复承诺
+
+## 常见违规借口与反驳
+
+| 借口 | 反驳 |
+|------|------|
+| 「代码 diff 直接看就行了，不需要重建上下文」 | 仅看 diff 会遗漏需求边界、架构选择与迁移策略；`reconstruct` 必须先产出最低充分重建记录。 |
+| 「PR 没有关联任务，直接写个报告发出去」 | 默认阻塞并要求先关联 Issue/task；只有显式「仅一次性检视」才走不可恢复降级路径。 |
+| 「把完整重建过程也发到 PR 普通评论，方便贡献者看」 | 过程原文只同步到 Issue artifact 评论；PR 只保留正式 Review，避免重复远端副本与恢复源混淆。 |
+| 「审查完了顺手提交一下」 | 本技能绝不执行 `git add`/`git commit`；提交是用户显式发起的独立步骤。 |
+
+## 第 0 步：状态核对（执行前硬约束）
+
+在加载本技能与相关规则后、做任何任务状态判断或用户可见结论之前，必须先执行状态核对：
+
+```bash
+agent-infra-internal task-snapshot {task-id} --format text
+```
+
+任务锚定路径在开始前记录当前任务目录与既有 `pr-review*` 轮次；一次性路径（无任务）记录 PR 编号与目标目录。状态核对输出粘贴到本轮产物 `pr-review-rN.md` 的 `## 状态核对` 段。
+
+## 执行步骤
+
+### 1. 解析入参与宿主解析
+
+解析目标 PR 编号 `{pr-number}`（`--pr <number>`，或 PR URL，或省略时按当前分支反查）。执行：
+
+```bash
+agent-infra-internal platform-pr-review inspect --pr {pr-number} [--cwd <path>]
+agent-infra-internal pr-review-grade resolve-host --pr {pr-number} [--cwd <path>]
+```
+
+`resolve-host` 输出 `HostResolution`（`unique` / `ambiguous` / `none`），按以下策略分流：
+
+- **唯一宿主**：绑定 `{task-id}`，调用 `task-activity pr-review-inspect` 取得 canonical round、prepared/open 状态和最近成功审查身份，再按下述恢复规则开始本轮（进入步骤 2）。
+- **多宿主歧义**：`resolve-host` 返回 `ambiguous`，`decide` 会拒绝分类（fail closed）。立即停止并提示人工指定唯一宿主；不进入证据分类。
+- **无宿主**：默认阻塞，要求先建立 Issue/task 关联（给出关联指引），不自动创建/导入 Issue。仅当用户显式选择「仅一次性检视」时，过程文件写入 `.agents/workspace/reviews/{pr-number}/`（`recoverable: false`）。
+
+任务锚定路径在宿主与远端 head 确定后执行：
+
+```bash
+agent-infra-internal task-activity {task-id} pr-review-inspect
+```
+
+- `prepared`：复用该 artifact；`open` 且 head 未变：对同一 artifact/head 重放 start（no-op）并恢复该轮。
+- `open` 且 head 已变：先把旧 artifact 的正式 Review 状态写为 `superseded`，再用旧 artifact/head 执行 terminate；随后使用 inspect 返回的 next round。
+- 无 prepared/open：按 `reference/report-template.md` 创建 next artifact 骨架，先写状态核对、身份信息、被审 head 与 `正式 Review 状态: pending`。
+
+在上述分流完成后、步骤 2 证据分级开始前写 started：
+
+```bash
+agent-infra-internal task-activity {task-id} pr-review-start --agent {agent} \
+  --artifact {pr-review-artifact} --head {head-sha}
+```
+
+一次性路径不调用 `task-activity`。任意 started 后、正式 Review 明确尚未发布的受控失败，都先把 artifact 状态写为 `aborted`，再以同一 artifact/head 调用 `pr-review-terminate --outcome aborted --reason <single-line>`；发布结果不确定时保持 open，不臆断 aborted。
+
+### 2. 单次 decide：证据枚举 + 场景分类 + 风险分级 + 模式选择
+
+把宿主、artifact 存在性、head 状态与六个纯证据风险因素写入输入 JSON。prior review 只使用 inspect 返回的最高 `applied` / `no-op` artifact；pending/aborted/superseded/failed 只参与 round 连续性，不作为既有结论。调用一次：
+
+```bash
+agent-infra-internal pr-review-grade decide --input-file {decide-input.json} [--cwd <path>]
+```
+
+一次返回完整 `DecisionRecord`（`scenario` / `freshness` / `alignment` / `risk` / `mode` / `firstReview` / `reason`）。把决策输入与输出原样记录到 `pr-review-rN.md` 的「证据清单」段（AC3 可追溯）。`host.kind === 'ambiguous'` 时命令拒绝，回到步骤 1 的阻塞出口。
+
+### 3. 生成 `pr-review-rN.md`
+
+按 `reference/report-template.md` 完成本轮产物；任务锚定路径保留步骤 1 已写入的身份/head/pending 状态，不重新分配 round。`reconstruct`（或 `audit` 证据不足）时，在行级 finding 之前先写「重建上下文」段（需求边界 / 架构选择 / 影响面 / 验证覆盖，见 `reference/evidence-grading.md`）。frontmatter 或首部记录 `recoverable: true|false`（一次性路径为 `false`）。完成最终问题清单后，一次性冻结 `{verdict, blockers, major, minor}`；正式 Review 正文与步骤 6 的 complete payload 必须直接复用该对象，不从报告文案反向统计。
+
+### 4. 同步 Issue artifact 评论（任务锚定路径）
+
+依次同步 task 评论与 artifact 评论，取得稳定 comment URL：
+
+```bash
+agent-infra-internal platform-comment sync {task-id} --kind task --agent {agent}
+agent-infra-internal platform-comment sync {task-id} --kind artifact --artifact {pr-review-artifact} --agent {agent}
+```
+
+一次性路径（无任务/Issue）跳过本步。
+
+### 5. 核对 head 并发布正式 Review
+
+再次读取远端 head，确认未被审查期间漂移：
+
+```bash
+agent-infra-internal platform-pr-review inspect --pr {pr-number} [--cwd <path>]
+```
+
+若 head 与步骤 1 记录一致，组装正文（head SHA / 结论 / finding / receipt / Issue artifact 链接，**不含 marker**），发布正式 Review：
+
+```bash
+agent-infra-internal platform-pr-review publish --pr {pr-number} --scope {taskId|pr{pr-number}} --round {round} \
+  --commit {head-sha} --event {COMMENT|APPROVE|REQUEST_CHANGES} --body-file {review-body.md} [--dry-run] [--cwd <path>]
+```
+
+`publish` 由 core 生成并校验 marker（首行），按 marker + commit 幂等（重放 no-op；marker 命中但 commit 不一致稳定失败）。head 漂移时，先把旧 artifact 状态写为 `superseded`，再闭合旧轮：
+
+```bash
+agent-infra-internal task-activity {task-id} pr-review-terminate --agent {agent} \
+  --artifact {pr-review-artifact} --head {head-sha} \
+  --outcome superseded --reason "head changed before publish"
+```
+
+闭合成功后重新执行步骤 1，进入下一 canonical round；不得先写新轮 started。若 publish 调用的远端结果不确定，保留 open 并在重试时依赖 marker + commit 恢复。
+
+### 6. 回写发布结果与任务状态
+
+- **任务锚定路径**：publish 返回 `applied` / `no-op` 后，把 Review ID/URL 与同名正式 Review 状态写入 `pr-review-rN.md`「发布结果」段，再以步骤 3 冻结的同一组 verdict/counts 闭合活动日志：
+
+  ```bash
+  agent-infra-internal task-activity {task-id} pr-review-complete --agent {agent} \
+    --artifact {pr-review-artifact} --head {head-sha} --verdict {approved|changes-requested|commented} \
+    --blockers {blockers} --major {major} --minor {minor}
+  ```
+
+  `task-activity` 由 typed payload 生成 `Verdict: <结论>, blockers: N, major: N, minor: N → {pr-review-artifact}`，并经 `writeTask` 原子完成 Activity Log、`## 审查反馈` 链接与版本戳刷新；不改 `current_step`，也不把 receipt/head/Review URL 写入 NOTE。
+
+- **一次性路径（无任务）**：Review ID/URL 只写 `pr-review-rN.md`「发布结果」段，不调用 `task-activity`。
+
+### 7. 发布后回写再同步（任务锚定路径）
+
+步骤 6 已改写本地 `pr-review-rN.md`「发布结果」段与 task.md（活动日志），而步骤 4 同步的是旧快照。`verify_comment_content` / `verify_task_comment_content` 会全文比对，必须先再同步使本地与远端一致。依次调用：
+
+```bash
+agent-infra-internal platform-comment sync {task-id} --kind task --agent {agent}
+agent-infra-internal platform-comment sync {task-id} --kind artifact --artifact {pr-review-artifact} --agent {agent}
+```
+
+`platform-comment sync` 对既有 marker 评论是幂等原地更新，comment URL 不变。一次性路径无任务，跳过本步。
+
+### 8. 完成校验
+
+- **任务锚定路径**：
+
+  ```bash
+  agent-infra-internal task-verify {task-id} review-pr.completed --artifact {pr-review-artifact} --format text
+  ```
+
+- **一次性路径**：
+
+  ```bash
+  agent-infra-internal pr-review-grade verify-artifact --artifact-file {pr-review-artifact} [--cwd <path>]
+  ```
+
+退出码 0 通过；1 按输出修复后重跑；2 停止并告知用户需要人工介入。
+
+### 9. 告知用户
+
+按 `reference/output-guidance.md` 选择唯一出口（正常发布 / 阻塞要求关联 / 一次性检视），渲染下一步前读取 `.agents/rules/next-step-output.md`。
+
+## 完成检查清单
+
+- [ ] 已按证据分级完成 PR 审查并产出 `pr-review-rN.md`
+- [ ] 正式 Review 已发布到目标 PR，绑定被审 head SHA
+- [ ] 任务锚定路径的 Issue artifact/task 评论已同步并再同步
+- [ ] 已完成校验（`task-verify` 或 `verify-artifact`）
+- [ ] 已更新 task.md 并追加 Activity Log（任务锚定路径）
+
+## 停止
+
+完成检查清单后立即停止。不要自动提交。

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

@@ -0,0 +1,37 @@
+{
+  "skill": "review-pr",
+  "checks": {
+    "artifact": {
+      "file_pattern": "pr-review.md|pr-review-r{N}.md",
+      "required_sections": [
+        "Identity",
+        "Evidence List",
+        "Coverage Matrix",
+        "Findings",
+        "Publication Result",
+        "State Check",
+        "Evidence"
+      ],
+      "required_patterns": [
+        "^\\$ ",
+        "^- \\*\\*Reviewed Head SHA\\*\\*[:：]\\s*[0-9a-f]{40}\\s*$",
+        "^- \\*\\*Review Mode\\*\\*[:：]\\s*(?:verify|audit|reconstruct)\\s*$",
+        "^- \\*\\*Evidence Scenario\\*\\*[:：]\\s*S[123]\\s*$",
+        "^- \\*\\*receipt\\*\\*[:：]\\s*\\S"
+      ]
+    },
+    "activity-log": {
+      "expected_action_pattern": "Review PR \\(Round \\d+\\)"
+    },
+    "platform-sync": {
+      "when": "issue_number_exists",
+      "expected_comment_marker": "<!-- sync-issue:{task-id}:{artifact-stem} -->",
+      "verify_comment_content": true,
+      "verify_task_comment_content": true,
+      "verify_issue_type": false,
+      "verify_issue_fields": false,
+      "verify_milestone": false,
+      "expected_comment_marker_key": "artifact"
+    }
+  }
+}

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

@@ -0,0 +1,37 @@
+{
+  "skill": "review-pr",
+  "checks": {
+    "artifact": {
+      "file_pattern": "pr-review.md|pr-review-r{N}.md",
+      "required_sections": [
+        "身份信息",
+        "证据清单",
+        "覆盖矩阵",
+        "问题清单",
+        "发布结果",
+        "状态核对",
+        "证据原文"
+      ],
+      "required_patterns": [
+        "^\\$ ",
+        "^- \\*\\*被审 head SHA\\*\\*[:：]\\s*[0-9a-f]{40}\\s*$",
+        "^- \\*\\*审查模式\\*\\*[:：]\\s*(?:verify|audit|reconstruct)\\s*$",
+        "^- \\*\\*证据场景\\*\\*[:：]\\s*S[123]\\s*$",
+        "^- \\*\\*receipt\\*\\*[:：]\\s*\\S"
+      ]
+    },
+    "activity-log": {
+      "expected_action_pattern": "Review PR \\(Round \\d+\\)"
+    },
+    "platform-sync": {
+      "when": "issue_number_exists",
+      "expected_comment_marker": "<!-- sync-issue:{task-id}:{artifact-stem} -->",
+      "verify_comment_content": true,
+      "verify_task_comment_content": true,
+      "verify_issue_type": false,
+      "verify_issue_fields": false,
+      "verify_milestone": false,
+      "expected_comment_marker_key": "artifact"
+    }
+  }
+}

package/templates/.agents/skills/review-pr/reference/evidence-grading.en.md

@@ -0,0 +1,65 @@
+# Evidence Grading Reference
+
+This file fixes the evidence-graded decision flow of `review-pr`. All determinations are executed mechanically by the typed-core pure functions (`lib/pr-review/evidence-grading.ts`); the skill's prompt layer only orchestrates. This file explains the criteria and rationale; it does not replace the core.
+
+## Main Decision Flow
+
+```text
+Resolve PR base/head and linked Issue/task (host resolution)
+   ├─ Unique host → bind the task, enumerate artifact presence → evidence classification
+   ├─ Ambiguous hosts → fail closed (decide refuses to classify; require a human-pinned host)
+   └─ No host → block by default and require linking; explicit one-shot → reviews/{pr-number}/
+        ↓
+Evidence scenario classification (S1/S2/S3) → freshness/alignment → risk grading → mode selection
+```
+
+## Host Resolution
+
+- Exactly one hit → `unique`; multiple without a unique winner → `ambiguous` (candidate list); zero hits → `none`.
+- Ambiguous hosts fail closed and never enter evidence classification. `pr-review-grade resolve-host --pr <N>` returns a typed `HostResolution`, and `decide` explicitly rejects `ambiguous`.
+- `Closes/Fixes #N` in the PR body (case-insensitive, comma/space separated, lists supported) is parsed by `extractClosingIssueNumbers`; a local `active/*/task.md` `pr_number` hit takes priority over the `issue_number` reverse lookup, and a task hit via both paths deduplicates to a single candidate.
+
+## Evidence Scenario Classification (S1 → S2 → S3)
+
+| Scenario | Criteria |
+|----------|----------|
+| S1 complete & trusted | unique task with matching issue_number + complete `analysis`/`plan`/`code` and all three review families + latest `pr-review*` reviewed head == current head + trusted source |
+| S2 partial / suspect | task exists but key artifacts are missing, or the head drifted, or the source is untrusted / cannot prove alignment with the current head |
+| S3 PR-only | no host; or unique task with no lifecycle artifacts at all and no prior `pr-review*` |
+
+- S3(b) symmetric criterion: `!hasAnalysis && !hasPlan && !hasCode && !hasReviewAnalysis && !hasReviewPlan && !hasReviewCode && !hasPriorPrReview`. A malformed task with only one review family (review-analysis / review-plan / review-code) falls to S2 → audit.
+- On first review (no prior `pr-review*`) S1(c) is unsatisfied, so a complete-lifecycle task falls to S2 → audit.
+
+## Freshness and Alignment
+
+- Freshness benchmark = the reviewed head SHA recorded by the latest `pr-review*`, compared character-for-character with the current head: match → `fresh`, otherwise → `stale`.
+- Alignment = freshness is `fresh` and the `issue_number` / `pr_number` match the resolution result.
+- No prior `pr-review*` → `n/a` / `n/a` (first-review special case).
+
+## Risk Grading (pure evidence)
+
+| Factor | LOW | HIGH |
+|--------|-----|------|
+| Change size | few files, small net additions | many files, large net additions |
+| Change sensitivity | no protected paths touched | touches lifecycle / rule / skill / config / security / auth / idempotency / concurrency paths |
+| Structural change | no schema/frontmatter/migration/interface-contract changes | involves schema / frontmatter / migration / interface contract |
+| Test coverage | tests present and passing | no tests or tests missing/failing |
+| Source credibility | a reviewable trusted lifecycle exists and aligns with the current head | no trusted record, or the source is not reviewable |
+| Recovery/merge impact | recovery contract unaffected | may break the recovery contract or create a dual-write source |
+
+Aggregation: any priority factor (sensitivity / source credibility) HIGH → HIGH; otherwise any HIGH → MEDIUM; all LOW → LOW. Identity factors never participate in any factor determination.
+
+## Mode Selection Matrix
+
+| Evidence scenario | Freshness/alignment | Risk | Mode |
+|-------------------|---------------------|------|------|
+| S1 | fresh + aligned | LOW/MEDIUM | verify (lightweight recheck) |
+| S1 | fresh + aligned | HIGH | audit (evidence audit) |
+| S2 | stale / misaligned / partial-missing | any | audit |
+| S3 | n/a | any | reconstruct (reconstruction review) |
+
+Running `review-pr` on a complete-lifecycle task for the first time falls to S2 → audit because there is no prior head record; "trusted lifecycle records → verify" corresponds to the re-review case where a prior `pr-review*` exists and the head has not drifted.
+
+## Minimum-Sufficient Reconstruction
+
+When the mode is `reconstruct` (or `audit` with insufficient evidence), form a reconstruction record before the line-level review covering at least: requirement boundary, architecture choices, impact surface, validation coverage; the on-disk order is "Reconstruction Context → Coverage Matrix → line-level Findings".