用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/marblo-app/marblo --skill tf-review命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
Use before merging agent-generated code — review a diff for correctness, security, and simplicity, and report findings ranked by severity.
Analyze requirements and map out components, roles, and dependencies.
Bulk-create tasks in Marblo MCP from the analysis results.
基于 SOC 职业分类
正在显示 SKILL.md
| name | tf-review |
| description | Review REVIEW-status tasks as PM, check code quality, then approve or reject. |
| disable-model-invocation | true |
| allowed-tools | Bash, Read, Glob, Grep |
Review the REVIEW tasks agents submitted. Check code quality and approve (DONE) or reject (revert to TODO).
Never use the Claude Code built-in tools (TaskCreate, TaskList, TaskUpdate, TaskGet). Task lookup/status-change MUST use the Marblo MCP tools:
get_all_tasks,update_task_status,add_activity,get_task_activities
Filter get_all_tasks to only REVIEW status.
👀 Awaiting review: {n}
━━━━━━━━━━━━━
1. TASK-002: User API (backend)
Submitted: 30 min ago | Owner: backend-agent
2. TASK-004: URL input screen (frontend)
Submitted: 15 min ago | Owner: frontend-agent
For each task, check the following:
Use get_task_activities to understand what the agent did.
Read and review the files listed in the task's scope field.
□ Feature completeness: are all requirements from the task description implemented?
□ Code quality: did it follow the skill file's rules (framework, patterns)?
□ Tests included: is there test code and does it pass?
□ Error handling: are exceptional cases handled properly?
□ Scope compliance: did it avoid modifying files outside scope?
□ Security: no vulnerabilities like SQL injection, XSS?
Show the review result to the user:
📝 TASK-002: User API review result
━━━━━━━━━━━━━━━━━━━━━━━━━
✅ Feature completeness: signup, login, profile lookup — OK
✅ Code quality: FastAPI + Pydantic v2 — OK
✅ Tests: 5/5 pass — OK
⚠️ Error handling: no password validation
✅ Scope compliance: OK
Verdict: approve or reject?
update_task_status → DONE✅ TASK-002 approved → DONE
🔓 TASK-004, TASK-005 are now workable
update_task_status → TODO (include a feedback comment)add_activity:
❌ TASK-002 rejected → TODO
Feedback: "Add password validation (8+ chars, include special char).
Make error messages user-friendly too."
After handling all REVIEWs, show the status:
📊 Review complete:
Approved: {n}
Rejected: {n}
Newly available tasks: {list}
Next action: /tf-work or re-run the agents