AGENT SKILL
repo-verify-hard
Executable — discover harness hard skills; run verify or fallback static/test/harness/build chain.
- License 许可
- MIT
- 所属仓库
- midyan/skissue
- 最近更新
- 2026-05-18
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- repo-verify-hard
- description
- Executable — discover harness hard skills; run verify or fallback static/test/harness/build chain.
02 资源文件 6
- 其他.no-auto-run—
- 参考文档INDEX.md950 B
- 脚本index.test.ts14.6 KB
- 脚本index.ts5.4 KB
- 参考文档README.md1.3 KB
- 脚本run-plan.test.ts1.7 KB
03 SKILL.md 正文
repo-verify (hard/)
Contract
- Exit 0 — discovery printed; all executed steps succeeded (or dry-run / discover-only).
- Exit 1 — a subprocess failed, or no verification steps could be resolved.
Implementation notes
- Discovery reads
harness/<id>/hard/index.tspresence andhard/.no-auto-run(same semantics asharness/runner.ts). resolveVerifyPlanreturnsnpm run verifywhenpackage.jsonhasscripts.verify.- Pre-commit:
.husky/pre-commitinvokesnpm run verify(Husky); document in soft SKILL.md so consumers wire the same gate locally.
04 同仓库更多 Skills
validate-shell-exec
Restrict child_process.spawn/exec to approved modules (e.g. shell/tmux helpers) so subprocess usage stays centralized.
★ 0
validate-registry-hard
Validates registry.json shape and that each skill path contains SKILL.md at the skill root.
★ 0
validate-registry
Keep root registry.json consistent with on-disk skill folders so installs and docs do not drift.
★ 0
validate-naming
Enforce kebab-case filenames and PascalCase type names under src/; reports naming drift for consistency.
★ 0
validate-logging
Prefer structured logging over raw console.* outside approved CLI entrypoints; keeps observability consistent.
★ 0
validate-links
Scan markdown for broken relative links across the repo. Run via npm run check:links or the harness hard skill.
★ 0