菜单
Architecture design + review loop for the aiagent project. The architect designs a solution following DDD patterns, then the architect-reviewer scores it against the 6 pillars. Loops until score >= 8/10 or 3 iterations maximum.
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。
基于 SOC 职业分类
Full Agent Development Life Cycle pipeline for the aiagent project. Runs story refinement → design → implementation planning (with human approval checkpoint) → implementation → testing → security → database → devops. Each coding phase uses a doer + antagonistic reviewer loop requiring score >= 8/10.
Create a detailed user story for the aiagent project with acceptance criteria and technical breakdown. Uses the story-refiner agent to explore the codebase and produce a structured story artifact.
Database schema + review loop for the aiagent project. The dba agent designs schema changes ensuring JSON/MongoDB storage duality, then the dba-reviewer scores the changes against the 6 pillars. Loops until score >= 8/10 or 3 iterations maximum.
Deployment readiness check for the aiagent project. Verifies the release pipeline, Docker build, and all 7 binary targets are ready before tagging a release. Uses the devops agent.
DevOps pipeline + review loop for the aiagent project. The devops agent manages CI/CD and Docker changes, then the devops-reviewer scores against the 6 pillars. Loops until score >= 8/10 or 3 iterations maximum.
Code implementation + review loop for the aiagent project. The developer implements changes following the approved plan, then the developer-reviewer scores the code against the 6 pillars including a mandatory duplication check. Loops until score >= 8/10 or 3 iterations maximum.
| name | design-review |
| description | Architecture design + review loop for the aiagent project. The architect designs a solution following DDD patterns, then the architect-reviewer scores it against the 6 pillars. Loops until score >= 8/10 or 3 iterations maximum. |
| user-invocable | true |
| argument-hint | add a rate-limiting layer to AI provider calls |
| context | fork |
Design and review the following:
You are orchestrating an Architecture Design Review cycle for the aiagent project.
Phase 1: Design
subagent_type: "architect" to design the solutionPhase 2: Review
3. Use the Agent tool with subagent_type: "architect-reviewer" to review the design
4. Pass the architect's full design document to the reviewer
5. The reviewer will score against the 6 pillars (Maintainability, Reliability, Scalability, Usability, Security, Quality)
Phase 3: Iterate 6. If overall score >= 8/10 (APPROVED): Report the approved design and scores 7. If overall score < 8/10 (REVISE):
Phase 4: Report 8. Present:
/adlc implementation planning