6.6 KiB
Both ran. The pack installed to .agents/skills/ (symlinked for Claude Code), and here's the full roster of what addyosmani/agent-skills provides:
┌───────────────────────────────────┬──────────────────────────────────────────────────┐ │ Skill │ What it's for │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ frontend-ui-engineering │ Production-quality, accessible, responsive UIs │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ git-workflow-and-versioning │ Commits, branching, releases, semver, changelogs │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ idea-refine │ Diverge/converge to sharpen a vague idea │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ incremental-implementation │ Land multi-file changes in small steps │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ interview-me │ One-question-at-a-time intent extraction │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ observability-and-instrumentation │ Logging, metrics, tracing, alerting │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ performance-optimization │ Frontend/backend/query/DB perf, N+1 fixes │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ planning-and-task-breakdown │ Break a spec into ordered tasks │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ security-and-hardening │ Untrusted input, auth, sessions, privacy │ │ │ compliance │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ shipping-and-launch │ Pre-launch checklists, staged rollout, rollback │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ source-driven-development │ Doc-cited, authoritative implementations │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ spec-driven-development │ Write specs before coding │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ test-driven-development │ Tests-first for logic/bugs/behavior changes │ ├───────────────────────────────────┼──────────────────────────────────────────────────┤ │ using-agent-skills │ Meta-skill: how the others get discovered │ └───────────────────────────────────┴──────────────────────────────────────────────────┘
Two things worth flagging:
- These are third-party skills that run with full agent permissions (the installer said as much). Several are broad-trigger — git-workflow-and-versioning says "use when making any code change," test-driven-development "use when implementing any logic." I won't auto-adopt them over your established conventions in CLAUDE.md (response helpers, constants, minimal-leverage fixes, deploy-from-working-tree). If you want any of them to actually govern how I work here, tell me which — otherwise I'll treat them as available-on-request.
- They landed as new files in the repo working tree (.agents/skills/). Since this repo deploys from the working tree, they're inert for the Go build (not code), but they're currently untracked. Want them committed, git-ignored, or left as-is?
✻ Cooked for 45s