@fitlab-ai/agent-infra 0.4.1 → 0.4.3

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

@@ -54,12 +54,59 @@ description: "处理代码审查反馈并修复问题"
 date "+%Y-%m-%d %H:%M:%S"
 ```
 
-更新 task.md，并追加：
-`- {yyyy-MM-dd HH:mm:ss} — **Refinement (Round {N}, for {review-artifact})** by {agent} — Fixed {n} blockers, {n} major, {n} minor issues → {refinement-artifact}`
+更新 task.md：
+- 审查 `## 需求` 段落，仅把因本轮修复而新满足且有测试通过支撑的条目从 `- [ ]` 勾为 `- [x]`
+- 追加：
+  `- {yyyy-MM-dd HH:mm:ss} — **Refinement (Round {N}, for {review-artifact})** by {agent} — Fixed {n} blockers, {n} major, {n} minor issues → {refinement-artifact}`
 
-### 7. 告知用户
+如果 task.md 中存在有效的 `issue_number`，执行以下同步操作（任一失败则跳过并继续）：
+- 执行前先读取 `.agents/rules/issue-sync.md`
+- 设置 `status: in-progress`
+- 按 `.agents/rules/issue-sync.md` 的 `in:` label 同步规则，基于分支改动精修 `in:` label（有映射时可增可删，无映射时仅补充）
+- 同步 `## 需求` 中已勾选项到 Issue body
+- 发布 `{refinement-artifact}` 评论
+- 创建或更新 `<!-- sync-issue:{task-id}:task -->` 评论（按 issue-sync.md 的 task.md 评论同步规则）
 
-如果本轮修复了 Blocker 或 Major，默认推荐重新审查。只有在风险确实很低且只修复 Minor 时，才可把直接提交作为备选项。
+### 7. 完成校验
+
+运行完成校验，确认任务产物和同步状态符合规范：
+
+```bash
+node .agents/scripts/validate-artifact.js gate refine-task .agents/workspace/active/{task-id} {refinement-artifact} --format text
+```
+
+处理结果：
+- 退出码 0（全部通过）-> 继续到「告知用户」步骤
+- 退出码 1（校验失败）-> 根据输出修复问题后重新运行校验
+- 退出码 2（网络中断）-> 停止执行并告知用户需要人工介入
+
+将校验输出保留在回复中作为当次验证输出。没有当次校验输出，不得声明完成。
+
+### 8. 告知用户
+
+> 仅在校验通过后执行本步骤。
+
+> **重要**：以下「下一步」中列出的所有 TUI 命令格式必须完整输出，不要只展示当前 AI 代理对应的格式。
+
+输出修复摘要后，根据修复的问题严重程度选择下一步：
+
+如果修复了 Blocker 或 Major（推荐重新审查）：
+
+```
+下一步 - 代码审查：
+  - Claude Code / OpenCode：/review-task {task-id}
+  - Gemini CLI：/agent-infra:review-task {task-id}
+  - Codex CLI：$review-task {task-id}
+```
+
+如果仅修复 Minor（可直接提交）：
+
+```
+下一步 - 提交代码：
+  - Claude Code / OpenCode：/commit
+  - Gemini CLI：/agent-infra:commit
+  - Codex CLI：$commit
+```
 
 ## 完成检查清单
 

package/templates/.agents/skills/refine-task/config/verify.json

@@ -0,0 +1,37 @@
+{
+  "skill": "refine-task",
+  "checks": {
+    "task-meta": {
+      "required_fields": [
+        "id",
+        "type",
+        "workflow",
+        "status",
+        "created_at",
+        "updated_at",
+        "current_step",
+        "assigned_to"
+      ]
+    },
+    "artifact": {
+      "file_pattern": "refinement.md|refinement-r{N}.md",
+      "required_sections": [
+        "审查反馈处理",
+        "修复后的测试结果"
+      ],
+      "freshness_minutes": 30
+    },
+    "activity-log": {
+      "expected_action_pattern": "Refinement \\(Round \\d+, for .+\\)",
+      "freshness_minutes": 30
+    },
+    "github-sync": {
+      "when": "issue_number_exists",
+      "expected_status_label": "status: in-progress",
+      "expected_comment_marker": "<!-- sync-issue:{task-id}:{artifact-stem} -->",
+      "verify_comment_content": true,
+      "verify_task_comment_content": true,
+      "sync_checked_requirements": true
+    }
+  }
+}

package/templates/.agents/skills/refine-title/SKILL.md

@@ -37,7 +37,7 @@ Based on the fetched data:
 
 **Generate Subject**:
 - **Ignore the original title** (avoid bias) - extract core intent from body
-- Keep concise (under 50 characters), English imperative mood, no trailing period
+- Keep concise (under 50 characters), use the content's original language (Chinese content stays Chinese, English content stays English), no trailing period
 
 ### 3. Present Suggestion
 
@@ -68,10 +68,18 @@ gh issue edit <id> --title "<new-title>"
 gh pr edit <id> --title "<new-title>"
 ```
 
+### 5. Inform User
+
+> **IMPORTANT**: All TUI command formats listed below must be output in full. Do not show only the format for the current AI agent.
+
+If the skill updated an Issue title, explain that no extra sync command is required; continue with the workflow skill that matches the task's current stage.
+
+If the skill updated a PR title, explain that `create-pr` now publishes the reviewer summary inline, so no extra sync command is needed; continue with the workflow skill that matches the task's current stage.
+
 ## Advantages
 
 This skill:
-1. **Fixes misleading titles**: Even if the original title is "Help me", it reads the body and generates a proper title like `fix(core): resolve startup error`
+1. **Fixes misleading titles**: Even if the original title is "Help me", it reads the body and generates a proper title like `fix(core): resolve startup error` while keeping the subject in the content's original language
 2. **Accurate scope**: By analyzing PR file changes, it can automatically infer the correct scope without manual specification
 
 ## Notes

package/templates/.agents/skills/refine-title/SKILL.zh-CN.md

@@ -37,7 +37,7 @@ gh pr view <id> --json number,title,body,labels,state,files
 
 **生成 Subject**：
 - **忽略原始标题**（避免偏见）- 从 body 中提取核心意图
-- 保持简洁（不超过 50 字符），英文祈使语气，末尾无句号
+- 保持简洁（不超过 50 字符），使用内容原始语言（中文内容用中文，英文内容用英文），末尾无句号
 
 ### 3. 展示建议
 
@@ -68,10 +68,18 @@ gh issue edit <id> --title "<new-title>"
 gh pr edit <id> --title "<new-title>"
 ```
 
+### 5. 告知用户
+
+> **重要**：以下「下一步」中列出的所有 TUI 命令格式必须完整输出，不要只展示当前 AI 代理对应的格式。
+
+如果修改了 Issue 标题，提示无需额外同步命令；后续按任务当前阶段继续执行对应工作流技能。
+
+如果修改了 PR 标题，提示 `create-pr` 已内联发布 reviewer 摘要，无需额外同步命令；后续按任务当前阶段继续执行对应工作流技能。
+
 ## 优势
 
 本技能的优势：
-1. **修复误导性标题**：即使原始标题是"Help me"，也能读取 body 并生成合适的标题，如 `fix(core): resolve startup error`
+1. **修复误导性标题**：即使原始标题是"Help me"，也能读取 body 并生成合适的标题，如 `fix(core): 修复启动错误` 或 `fix(core): resolve startup error`
 2. **精确 scope**：通过分析 PR 文件变更，可以自动推断正确的 scope，无需手动指定
 
 ## 注意事项

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

@@ -0,0 +1,159 @@
+---
+name: restore-task
+description: "Restore local task files from GitHub Issue comments"
+---
+
+# Restore Task
+
+Restore local task workspace files from GitHub Issue comments that contain sync markers.
+
+## Boundary / Critical Rules
+
+- Restore files only from comments marked with `<!-- sync-issue:{task-id}:... -->`
+- Restore into `.agents/workspace/active/{task-id}/` by default
+- Stop immediately if the target directory already exists and ask the user to resolve the conflict first
+- After executing this skill, you **must** immediately update the restored `task.md`
+
+## Steps
+
+### 1. Verify Input and Environment
+
+Check:
+- required `{issue-number}`
+- optional `{task-id}`
+- `gh auth status`
+
+If the user provided `{task-id}`, validate the `TASK-{yyyyMMdd-HHmmss}` format.
+
+### 2. Fetch Issue Comments
+
+Read all Issue comments while preserving the original order and comment IDs.
+
+Suggested command:
+
+```bash
+gh api "repos/{owner}/{repo}/issues/{issue-number}/comments" --paginate
+```
+
+### 3. Determine the task-id and Files to Restore
+
+Filter comments by these hidden markers:
+
+```html
+<!-- sync-issue:{task-id}:{file-stem} -->
+<!-- sync-issue:{task-id}:{file-stem}:{part}/{total} -->
+```
+
+Rules:
+- when `{task-id}` was provided, match only that task
+- when `{task-id}` was omitted, infer it from the `<!-- sync-issue:{task-id}:task -->` comment first
+- if you cannot determine a unique task-id, stop and tell the user
+- ignore `summary` marker comments because they are complete-task aggregate output rather than restorable local task files
+- map `{file-stem}` back to filenames:
+  - `task` -> `task.md`
+  - `analysis` / `analysis-r{N}` -> matching `.md`
+  - `plan` / `plan-r{N}` -> matching `.md`
+  - `implementation` / `implementation-r{N}` -> matching `.md`
+  - `review` / `review-r{N}` -> matching `.md`
+  - `refinement` / `refinement-r{N}` -> matching `.md`
+
+### 4. Process Chunks and Check the Local Directory
+
+Read `.agents/rules/issue-sync.md` before executing this step.
+
+For each file:
+- collect its single comment or chunked comments
+- for `task.md` comments, reverse the `<details>` frontmatter wrapper described in issue-sync.md before reassembling the file body
+- when `{part}/{total}` exists, sort by part and verify the set is complete
+- extract the file body by removing the hidden marker, heading, and footer
+- concatenate chunk bodies into the final file content
+
+Before writing any file, verify that:
+- `.agents/workspace/active/{task-id}/` does not exist
+
+If the directory already exists, stop immediately and tell the user to handle it manually first.
+
+### 5. Write the Local Files
+
+Create `.agents/workspace/active/{task-id}/` and write files back in this order:
+
+1. `task.md`
+2. every other restored artifact file in filename order
+
+Write only files that were actually recovered from Issue comments. Do not invent missing files.
+
+### 6. Update the Restored task.md
+
+Get the current time:
+
+```bash
+date "+%Y-%m-%d %H:%M:%S"
+```
+
+Update the restored `task.md`:
+- `status`: `active`
+- `assigned_to`: {current AI agent}
+- `updated_at`: {current time}
+- keep the original `current_step`
+- append this entry to `## Activity Log`:
+  ```
+  - {yyyy-MM-dd HH:mm:ss} — **Restore Task** by {agent} — Restored task from Issue #{issue-number}
+  ```
+
+### 7. Verification Gate
+
+Run the verification gate:
+
+```bash
+node .agents/scripts/validate-artifact.js gate restore-task .agents/workspace/active/{task-id} --format text
+```
+
+Handle the result as follows:
+- exit code 0 (all checks passed) -> continue to the "Inform User" step
+- exit code 1 (validation failed) -> fix the reported issues and run the gate again
+- exit code 2 (network blocked) -> stop and tell the user that human intervention is required
+
+Keep the gate output in your reply as fresh evidence. Do not claim completion without output from this run.
+
+### 8. Inform User
+
+> Execute this step only after the verification gate passes.
+
+> **IMPORTANT**: All TUI command formats listed below must be output in full. Do not show only the format for the current AI agent.
+
+Output format:
+
+```text
+Task {task-id} was restored from Issue #{issue-number}.
+
+Summary:
+- Restored files: {count}
+- Task directory: .agents/workspace/active/{task-id}/
+- Current step: {current_step}
+
+Next step - check task status:
+  - Claude Code / OpenCode: /check-task {task-id}
+  - Gemini CLI: /{{project}}:check-task {task-id}
+  - Codex CLI: $check-task {task-id}
+```
+
+## Completion Checklist
+
+- [ ] Fetched and parsed Issue comments
+- [ ] Restored `task.md` and every available artifact file
+- [ ] Updated the restored task.md
+- [ ] Ran and passed the verification gate
+- [ ] Showed the next-step commands in every TUI format
+
+## STOP
+
+Stop after completing the checklist. Do not continue the workflow automatically.
+
+## Error Handling
+
+- Issue missing or inaccessible
+- `gh` unavailable or unauthenticated
+- No sync-marked comments found
+- Unable to determine a unique `task-id`
+- Target directory already exists
+- Missing chunks or incomplete chunk ordering

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

@@ -0,0 +1,159 @@
+---
+name: restore-task
+description: "从 GitHub Issue 评论还原本地任务文件"
+---
+
+# 还原任务
+
+从带有 sync 标记的 GitHub Issue 评论中恢复本地任务工作区文件。
+
+## 行为边界 / 关键规则
+
+- 只从带 `<!-- sync-issue:{task-id}:... -->` 标记的评论恢复文件
+- 默认恢复到 `.agents/workspace/active/{task-id}/`
+- 如果目标目录已存在，立即停止并提示用户先处理目录冲突
+- 执行本技能后，你**必须**立即更新恢复出的 `task.md`
+
+## 执行步骤
+
+### 1. 验证输入与环境
+
+检查：
+- 必填参数 `{issue-number}`
+- 可选参数 `{task-id}`
+- `gh auth status`
+
+如果用户传入了 `{task-id}`，校验其格式为 `TASK-{yyyyMMdd-HHmmss}`。
+
+### 2. 获取 Issue 评论
+
+读取 Issue 的全部评论，保留原始顺序和评论 ID。
+
+建议命令：
+
+```bash
+gh api "repos/{owner}/{repo}/issues/{issue-number}/comments" --paginate
+```
+
+### 3. 确定 task-id 与待恢复文件
+
+从评论中筛选隐藏标记：
+
+```html
+<!-- sync-issue:{task-id}:{file-stem} -->
+<!-- sync-issue:{task-id}:{file-stem}:{part}/{total} -->
+```
+
+处理规则：
+- 用户提供了 `{task-id}` 时，仅匹配该任务
+- 未提供时，优先从 `<!-- sync-issue:{task-id}:task -->` 评论推断
+- 若找不到唯一 task-id，立即停止并告知用户
+- 忽略 `summary` 标记评论；它是 complete-task 的聚合产物，不对应本地任务文件
+- 将 `{file-stem}` 映射回文件名：
+  - `task` -> `task.md`
+  - `analysis` / `analysis-r{N}` -> 对应 `.md`
+  - `plan` / `plan-r{N}` -> 对应 `.md`
+  - `implementation` / `implementation-r{N}` -> 对应 `.md`
+  - `review` / `review-r{N}` -> 对应 `.md`
+  - `refinement` / `refinement-r{N}` -> 对应 `.md`
+
+### 4. 处理分片并检查本地目录
+
+执行本步骤前先读取 `.agents/rules/issue-sync.md`。
+
+对每个文件执行：
+- 收集单条评论或分片评论
+- 对 `task.md` 评论按 issue-sync.md 中的 `<details>` frontmatter 格式反向拆解，提取 frontmatter 后再与正文拼合
+- 如存在 `{part}/{total}`，按 part 升序排序并校验分片完整
+- 从评论正文中提取文件内容，去掉隐藏标记、标题和页脚
+- 拼接得到最终文件内容
+
+在写文件前检查：
+- `.agents/workspace/active/{task-id}/` 不存在
+
+如果目录已存在，立即停止并提示用户先手动处理。
+
+### 5. 写回本地文件
+
+创建 `.agents/workspace/active/{task-id}/`，按以下顺序写回：
+
+1. `task.md`
+2. 其余产物文件（按文件名排序）
+
+仅写回从 Issue 评论中实际恢复出的文件，不补造缺失文件。
+
+### 6. 更新恢复后的 task.md
+
+获取当前时间：
+
+```bash
+date "+%Y-%m-%d %H:%M:%S"
+```
+
+更新恢复出的 `task.md`：
+- `status`：`active`
+- `assigned_to`：{当前 AI 代理}
+- `updated_at`：{当前时间}
+- 保留原 `current_step`
+- 在 `## 活动日志` 追加：
+  ```
+  - {yyyy-MM-dd HH:mm:ss} — **Restore Task** by {agent} — Restored task from Issue #{issue-number}
+  ```
+
+### 7. 完成校验
+
+运行完成校验：
+
+```bash
+node .agents/scripts/validate-artifact.js gate restore-task .agents/workspace/active/{task-id} --format text
+```
+
+处理结果：
+- 退出码 0（全部通过）-> 继续到「告知用户」步骤
+- 退出码 1（校验失败）-> 根据输出修复问题后重新运行校验
+- 退出码 2（网络中断）-> 停止执行并告知用户需要人工介入
+
+将校验输出保留在回复中作为当次验证输出。没有当次校验输出，不得声明完成。
+
+### 8. 告知用户
+
+> 仅在校验通过后执行本步骤。
+
+> **重要**：以下「下一步」中列出的所有 TUI 命令格式必须完整输出，不要只展示当前 AI 代理对应的格式。
+
+输出格式：
+
+```text
+任务 {task-id} 已从 Issue #{issue-number} 还原。
+
+摘要：
+- 恢复文件：{数量}
+- 任务目录：.agents/workspace/active/{task-id}/
+- 当前步骤：{current_step}
+
+下一步 - 查看任务状态：
+  - Claude Code / OpenCode：/check-task {task-id}
+  - Gemini CLI：/{{project}}:check-task {task-id}
+  - Codex CLI：$check-task {task-id}
+```
+
+## 完成检查清单
+
+- [ ] 已获取并解析 Issue 评论
+- [ ] 已还原 `task.md` 和所有可用产物文件
+- [ ] 已更新恢复后的 task.md
+- [ ] 已运行并通过完成校验
+- [ ] 已向用户展示所有 TUI 格式的下一步命令
+
+## 停止
+
+完成检查清单后立即停止。不要自动继续执行工作流。
+
+## 错误处理
+
+- Issue 不存在或无权访问
+- `gh` 未认证
+- 找不到带 sync 标记的评论
+- 无法唯一确定 `task-id`
+- 目标目录已存在
+- 分片缺失或顺序不完整

package/templates/.agents/skills/restore-task/config/verify.json

@@ -0,0 +1,24 @@
+{
+  "skill": "restore-task",
+  "checks": {
+    "task-meta": {
+      "required_fields": [
+        "id",
+        "type",
+        "workflow",
+        "status",
+        "created_at",
+        "updated_at",
+        "current_step",
+        "assigned_to"
+      ],
+      "expected_status": "active",
+      "require_issue_number": true
+    },
+    "activity-log": {
+      "expected_action_pattern": "Restore Task",
+      "freshness_minutes": 30
+    },
+    "github-sync": null
+  }
+}

package/templates/.agents/skills/review-task/SKILL.md

@@ -53,7 +53,30 @@ date "+%Y-%m-%d %H:%M:%S"
 Update task.md and append:
 `- {yyyy-MM-dd HH:mm:ss} — **Code Review (Round {N})** by {agent} — Verdict: {Approved/Changes Requested/Rejected}, blockers: {n}, major: {n}, minor: {n} → {artifact-filename}`
 
-### 7. Inform User
+If task.md contains a valid `issue_number`, perform these sync actions (skip and continue on any failure):
+- Read `.agents/rules/issue-sync.md` before syncing
+- Set `status: in-progress`
+- Publish the `{review-artifact}` comment
+- Create or update the `<!-- sync-issue:{task-id}:task -->` comment (follow the task.md comment sync rule in issue-sync.md)
+
+### 7. Verification Gate
+
+Run the verification gate to confirm the task artifact and sync state are valid:
+
+```bash
+node .agents/scripts/validate-artifact.js gate review-task .agents/workspace/active/{task-id} {review-artifact} --format text
+```
+
+Handle the result as follows:
+- exit code 0 (all checks passed) -> continue to the "Inform User" step
+- exit code 1 (validation failed) -> fix the reported issues and run the gate again
+- exit code 2 (network blocked) -> stop and tell the user that human intervention is required
+
+Keep the gate output in your reply as fresh evidence. Do not claim completion without output from this run.
+
+### 8. Inform User
+
+> Execute this step only after the verification gate passes.
 
 Choose exactly one branch based on the findings:
 - no blockers, no major, no minor -> approved with no issues
@@ -71,6 +94,7 @@ Include all TUI command formats in the next-step output.
 - [ ] Created `{review-artifact}`
 - [ ] Updated task.md and appended the Activity Log entry
 - [ ] Chose exactly one verdict branch in the user output
+- [ ] Informed the user of the next step (must include all TUI command formats; do not filter)
 
 ## Notes
 

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

@@ -53,7 +53,30 @@ date "+%Y-%m-%d %H:%M:%S"
 更新 task.md，并追加：
 `- {yyyy-MM-dd HH:mm:ss} — **Code Review (Round {N})** by {agent} — Verdict: {Approved/Changes Requested/Rejected}, blockers: {n}, major: {n}, minor: {n} → {artifact-filename}`
 
-### 7. 告知用户
+如果 task.md 中存在有效的 `issue_number`，执行以下同步操作（任一失败则跳过并继续）：
+- 执行前先读取 `.agents/rules/issue-sync.md`
+- 设置 `status: in-progress`
+- 发布 `{review-artifact}` 评论
+- 创建或更新 `<!-- sync-issue:{task-id}:task -->` 评论（按 issue-sync.md 的 task.md 评论同步规则）
+
+### 7. 完成校验
+
+运行完成校验，确认任务产物和同步状态符合规范：
+
+```bash
+node .agents/scripts/validate-artifact.js gate review-task .agents/workspace/active/{task-id} {review-artifact} --format text
+```
+
+处理结果：
+- 退出码 0（全部通过）-> 继续到「告知用户」步骤
+- 退出码 1（校验失败）-> 根据输出修复问题后重新运行校验
+- 退出码 2（网络中断）-> 停止执行并告知用户需要人工介入
+
+将校验输出保留在回复中作为当次验证输出。没有当次校验输出，不得声明完成。
+
+### 8. 告知用户
+
+> 仅在校验通过后执行本步骤。
 
 必须先判断结果，再只选择一个输出分支：
 - 无 blocker、major、minor -> 通过且无问题
@@ -71,6 +94,7 @@ date "+%Y-%m-%d %H:%M:%S"
 - [ ] 已创建 `{review-artifact}`
 - [ ] 已更新 task.md 并追加 Activity Log
 - [ ] 用户输出中只选择了一个审查结论分支
+- [ ] 告知了用户下一步（必须展示所有 TUI 的命令格式，不要筛选）
 
 ## 注意事项
 

package/templates/.agents/skills/review-task/config/verify.json

@@ -0,0 +1,40 @@
+{
+  "skill": "review-task",
+  "checks": {
+    "task-meta": {
+      "required_fields": [
+        "id",
+        "type",
+        "workflow",
+        "status",
+        "created_at",
+        "updated_at",
+        "current_step",
+        "assigned_to"
+      ]
+    },
+    "artifact": {
+      "file_pattern": "review.md|review-r{N}.md",
+      "required_sections": [
+        "审查摘要",
+        "问题清单",
+        "结论与建议"
+      ],
+      "required_patterns": [
+        "^### 审查决定$"
+      ],
+      "freshness_minutes": 30
+    },
+    "activity-log": {
+      "expected_action_pattern": "Code Review \\(Round \\d+\\)",
+      "freshness_minutes": 30
+    },
+    "github-sync": {
+      "when": "issue_number_exists",
+      "expected_status_label": "status: in-progress",
+      "expected_comment_marker": "<!-- sync-issue:{task-id}:{artifact-stem} -->",
+      "verify_comment_content": true,
+      "verify_task_comment_content": true
+    }
+  }
+}

package/templates/.agents/skills/update-agent-infra/SKILL.md

@@ -129,4 +129,15 @@ Based on the script report and merged results, output a complete update report i
 - Ejected file processing
 - Self-update detection result
 
+If there are changes to submit, add:
+
+> **IMPORTANT**: All TUI command formats listed below must be output in full. Do not show only the format for the current AI agent.
+
+```
+Next step - commit changes:
+  - Claude Code / OpenCode: /commit
+  - Gemini CLI: /{{project}}:commit
+  - Codex CLI: $commit
+```
+
 Output the report, then **STOP** — do not make other changes to the project.

package/templates/.agents/skills/update-agent-infra/SKILL.zh-CN.md

@@ -119,4 +119,15 @@ node .agents/skills/update-agent-infra/scripts/sync-templates.js
 - ejected 文件处理
 - 自身更新检测结果
 
+如有变更需要提交，追加：
+
+> **重要**：以下「下一步」中列出的所有 TUI 命令格式必须完整输出，不要只展示当前 AI 代理对应的格式。
+
+```
+下一步 - 提交代码：
+  - Claude Code / OpenCode：/commit
+  - Gemini CLI：/agent-infra:commit
+  - Codex CLI：$commit
+```
+
 输出报告后**停止**，不要对项目做其他更改。