先研究、再确认、再规划。流程化的SKILL
SaferSkills independently audited workflow-skill (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
本 skill 的唯一设计来源是仓库根目录的 WORKFLOW.md。执行时必须优先遵守 WORKFLOW.md,并参考 BUG_REP.md 中记录的已知问题。不要把本文件理解成命令实现;它是 Claude/Codex 执行项目开发流程时的操作规约。
plan.md;如涉及 API 变动,还必须已有被用户认可的 api.md。task/ 模板与 WORKFLOW.md 不一致,以 WORKFLOW.md 为准,在 feature 工作副本中补全缺失章节。新建任务时:
feature/[YYYY]/。task/ 目录复制 research.md、plan.md、api.md 到 feature/[YYYY]/。feature/[YYYY]/[YYYYMMDD]-[task-name]-research.mdfeature/[YYYY]/[YYYYMMDD]-[task-name]-plan.mdfeature/[YYYY]/[YYYYMMDD]-[task-name]-api.md如果用户或项目已有命名规范,优先沿用已有规范,但必须保证三类文件可互相关联。
目标:填满 research 文件,验证 agent 是否真正理解代码库和需求。
必须完成:
task/research.md 缺失上述章节,在 feature 副本里按 WORKFLOW.md 补全。研究完成后必须停下,使用 AskUserQuestion 或等价方式向用户确认待确认问题。确认完成前,不得生成或填写 plan。
只有研究阶段被用户确认后,才能填写 plan 文件。
plan 文件必须包含:
标注规则:
task/plan.md 缺失上述章节,在 feature 副本里按 WORKFLOW.md 补全。在 plan 标注结束后,判断是否涉及 API 变动。
API 文档必须至少包含:
API 审阅规则:
执行只能在以下条件全部满足后开始:
执行方式:
plan-executor subagent。plan-executor,明确告知用户该限制;只有用户同意后,才由当前 agent 执行。推荐执行指令:
implement it all.
mark completed in plan.
do not stop until all done.
no unnecessary comments.
continuously run typecheck...执行完成后:
git revert,然后回到计划阶段重写方案。BUG_REP.md,确认是否有会影响本次流程的已知问题。BUG_REP.md,包含问题、触发条件、解决方案。BUG_REP.md 为空时,不要编造问题。~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.