Skip to content
5 changes: 4 additions & 1 deletion .github/skills/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -17,6 +17,7 @@ before push.
| Skill | Purpose |
|-------|---------|
| [`pr-review/`](pr-review/SKILL.md) | Multi-dimensional review of a PR / feature branch diff (security, correctness, API & DSL ergonomics, alternative solutions, test coverage, docs & samples sync, packaging/agent-kit impact, multi-model cross-check). Reports findings to stdout; does not apply fixes. |
| [`pr-lifecycle/`](pr-lifecycle/SKILL.md) | Take a PR to a clean `ready-for-review` state: run `pr-review` once and fix Critical/High/Medium findings, then loop GitHub Copilot review rounds until no meaningful findings remain. Every round requires green CI and a reply + resolve on each Copilot and `github-code-quality` thread. Tracks state with `agent-preparing` / `agent-blocked` / `ready-for-review` labels; waits via a session wakeup. Applies fixes and pushes; never merges. |
| [`perf-compare/`](perf-compare/SKILL.md) | Benchmark the `StressPerf.ReactorOptimized` data-grid harness on the current branch vs a clean `main` worktree (interleaved, same machine) and report a direction-aware delta for the four headline metrics. The local equivalent of commenting `/perf` on a PR. Reports to stdout; does not apply fixes. |
| [`coverage-uplift/`](coverage-uplift/SKILL.md) | Raise test coverage with non-vacuous tests: baseline via `tools/coverage`, classify each gap into the right tier (headless unit / selftest / E2E), write narrowly-scoped tests whose assertions fail if the target code is deleted, mutation-check the oracles, and cross-check with the pr-review multi-model dimension. Adds tests. |
| [`analyzer-dym/`](analyzer-dym/SKILL.md) | Author or remove Reactor "did-you-mean" / diagnostic analyzers and their `mur check` mirrors under the `netstandard2.0` analyzer constraints: FP-spike first, add the `AnalyzerReleases.Unshipped.md` row, keep CLI parity, sync generated docs. Applies changes. |
Expand All @@ -33,4 +34,6 @@ before push.
runnable code.
- Prompt fragments meant to be passed verbatim to sub-agents live under a
`dimensions/` (or similarly named) subfolder.
- Output goes to stdout unless the user explicitly asks for a file.
- Output goes to stdout unless the user explicitly asks for a file. The exception is
an authorized `pr-lifecycle` run, which also pushes fixes, replies to and resolves
review threads, and sets the PR's lifecycle label.
Loading
Loading