[Documentation] Use when generating, smart-merging or refactoring the root AI context (CLAUDE.md, AGENTS.md) and its Claude/Codex projections. Config file only: project-config.
中文简介正在准备中。当前显示原文说明。
duc01226/EasyPlatform122026年10月18日 更新
免费在 GitHub 公开
commit-review
[Code Quality] Review an exact pending commit when the commit skill selects lightweight review or the user requests it. --fix-loop repairs and re-reviews; no Git publication authority.
中文简介正在准备中。当前显示原文说明。
查看安装方法包含的文件（2）
- SKILL.md1.9 KB
- references/fix-loop.md1.8 KB
SKILL.md（原文）
安装前可以查看交给 AI 代理的指令内容。
- Resolve the intended commit and its purpose; load applicable project review rules and overlays. In
--fix-loop, read references/fix-loop.md and snapshot the exact candidate before review; preserve unrelated work. - Inspect every changed file and affected callers for correctness, regressions, security and unnecessary complexity. Trace changed behavior through success, failure and boundary cases; check permissions and affected callers where relevant. Check changed behavior against its configured canonical owner/spec, profile-declared scenario/case and executing test evidence; read further references only for an actual concern. For changed behavior, identify a realistic bug the test should catch; use a targeted mutation probe only when test strength is unclear and edits are authorized.
- Validate concerns inline against source and counterevidence. Findings need a reachable trigger, consequence, file:line, severity and ≥85% confidence; omit speculative advice, unrelated debt and issues already caught by checks. Missing required evidence blocks a pass.
- Default to read-only. With
--fix-loop, fix validated issues within scope and freshly review the whole updated candidate, at most two rounds;--loop-owner=callerstays read-only. Stop with unresolved MEDIUM+ or failed checks; defer nonblocking LOWs. Never extend the budget silently. - Report findings, fixes, checks, rounds and deferred LOWs briefly. Issue a receipt only after a fully covered fresh fix-loop pass and passing required checks, using its original snapshot. Never stage, commit or push from this review; do not load another review workflow routinely.
评价
还没有评价。欢迎分享你的使用感受。
评论
无需登录即可发表。欢迎分享使用感受、提问，或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
同一仓库的技能
[Code Quality] Use when a workflow step or the user asks for AI code or plan review: LLM calls, prompts, agents, RAG, evals and guardrails. --mode={code|plan}; --report-only is read-only.
中文简介正在准备中。当前显示原文说明。
duc01226/EasyPlatform122026年10月18日 更新
[Architecture] Use when a workflow step or the user asks for architecture design, compliance review, scalability grading or whole-project audit via --mode={design|review|scalability|full}.
中文简介正在准备中。当前显示原文说明。
duc01226/EasyPlatform122026年10月18日 更新
[Content] Use when a workflow step or the user asks for product/feature ideation with ranked opportunities and validation tests. --mode={roadmap|scope}.
中文简介正在准备中。当前显示原文说明。
duc01226/EasyPlatform122026年10月18日 更新
[Content] Use when a workflow step or the user asks for business viability: Business Model Canvas, financials, risks and go-to-market recommendation.
中文简介正在准备中。当前显示原文说明。
duc01226/EasyPlatform122026年10月18日 更新
[Code Quality] Use when a workflow step or the user asks for a focused code-change review: staged, unstaged, branch diff or simple pull request. --fix-loop fixes and re-reviews. Complex pull request: workflow-review-changes; rationale: why-review.
中文简介正在准备中。当前显示原文说明。
duc01226/EasyPlatform122026年10月18日 更新