A collection of reusable agent skills for software-engineering tasks.
Each top-level directory is a self-contained skill — drop it into the right place for your runtime:
- Copilot CLI — copy into
~/.agents/skills/<name>/(user-scoped) or<repo>/.agents/skills/<name>/(repository-scoped) - Claude Code — copy into
~/.claude/skills/<name>/or<repo>/.claude/skills/<name>/
All skills follow the standard layout:
<skill-name>/
├── SKILL.md # Front-matter (name, description) + instructions
├── references/ # Optional: deeper docs the agent loads on demand
└── scripts/ # Optional: helper scripts the skill invokes
Comprehensive reference for building and maintaining GitHub Agentic Workflows (gh-aw). Covers architecture, security boundaries, fork handling, safe outputs, anti-patterns, compilation, and troubleshooting. Targets the latest stable gh-aw release; migration tips for older workflows live in references/migrations.md.
Use it when creating or editing .github/workflows/*.md files, writing safe-outputs configs, configuring fork PR handling, setting up integrity filtering, or debugging "why doesn't my workflow trigger".
Scrapes the upstream github/gh-aw repo (releases, docs, commits) to detect what's changed since the gh-aw-guide skill was last updated, then helps refresh the guide. Only reports stable, non-prerelease releases — pre-release features should not be added to the guide until they ship in a stable release.
Use it when asked to "update gh-aw guide", "check for gh-aw changes", or "sync gh-aw knowledge".
Multi-model adversarial PR code review. Dispatches three independent reviewers (different models) in parallel, runs cross-validation on disputed findings, and produces a single high-signal review. Works with any GitHub repository — no project-specific knowledge required.
Use it when asked to "review a PR", "code review", or "expert review".
MIT — see each skill's SKILL.md front-matter for its own attribution.