AI Code Skills
A flat collection of Claude Code skills and supporting hooks for development, presentations, and content workflows.
These are the skills I (Antonio Sejas) use day-to-day across Automattic work and personal projects. They are intentionally small, single-purpose, and composable — drop in whichever ones you find useful.
Layout
skills/ # one folder per skill, each containing SKILL.md
hooks/ # standalone hook scripts grouped by purpose
No plugin manifest, no marketplace — just drop the folders where you want them.
Install
Copy or symlink any skill into ~/.claude/skills/:
ln -s "$PWD/skills/commit" ~/.claude/skills/commit
Or symlink the whole directory to expose every skill:
for d in skills/*/; do ln -s "$PWD/$d" "$HOME/.claude/skills/$(basename $d)"; done
Skills
Each skill is invoked by typing /<skill-name> in Claude Code, or by triggering a phrase from the skill's description (Claude auto-routes).
Development
| Skill | What it does | How I use it |
|---|
commit | Drafts a Conventional Commit message from the staged diff and creates the commit. | /commit when staging is ready. Lets me skip writing commit subjects by hand. |
explain | Locates a function in the codebase and explains it for a junior dev. | /explain handleAuth when onboarding to a new repo. |
worktree | Creates an isolated git worktree from origin/trunk in ~/worktrees.nosync/, copies node_modules. | /worktree feat/new-thing to start parallel feature work without polluting the main checkout. |
pr-review | Lists PRs awaiting my review, checks out each branch, runs the linter, and writes a structured review doc. | Daily review pass — pick one off the list and let the skill set up the workspace. |
quick-pr-review | Lightweight diff review against a GitHub PR URL (no checkout). | When I just need a fast read on a colleague's PR. |
ai-prs | Lists open PRs targeting the Studio AI command with status/approval/reviewers. | Triage Studio AI PRs before standup. |
react-doctor | Scans recent React changes for common issues (hook deps, key warnings, state misuse). | After finishing a React feature, before /commit. |
create-agents-md | Generates AGENTS.md / CLAUDE.md for a repository documenting conventions and structure. | When starting a new repo or onboarding a coding agent. |
studio-debug | Inspects WordPress Studio local sites — SQLite DB, logs, plugin/theme issues. | When a Studio site misbehaves locally. |
teach | Acts as a teacher that makes me deeply understand the current session — problem, solution, edge cases, broader impact — via a running checklist, incremental confirmation, and quizzes (AskUserQuestion). | /teach after Claude lands a non-trivial change I want to actually internalize, not just merge. |
score | Calibrated 0–10 judging of any artifact — website/UI (screenshot), markdown draft (direct read), or code diff. A fresh judge subagent reasons first (strengths/weaknesses/improvements/reasoning), score last, with calibration anchors and a pinned "what a 10 looks like" bar so improve-until-N loops converge instead of chasing moving goalposts. | /score the landing page, then a goal loop "repeat until 9/10" — the judge drives design/writing iteration. |
start-issue | From a Linear link or branch name: resolves the issue (Linear MCP, context-a8c fallback), spins up a worktree, implements the fix, verifies (lint/typecheck/tests, light+dark for UI), then commits, pushes, and opens a draft PR with the template filled in. | /start-issue <linear-link> to go from ticket to draft PR in one shot. |
unworktree | Reverse of worktree: absorbs a worktree's branch back into the main checkout — dirty-tree safety gate, removes the worktree, checks the branch out in the main repo. Set $MAIN_REPO to your main checkout. | When I'm done in a ~/worktrees.nosync/ branch and want to continue it in the main repo. |
explore-exploit | Generate-many-then-select for tasks with a large solution space and a measurable quality signal (naming, UI variants, copy, code/prompt tuning). Fans out candidates, scores them, and iterates toward the best — tournament / simulated-annealing style. | "Generate N options and rank them" or an iterate-until-score loop. |
Standup / Daily
'%20stop-opacity%3D'0.16'%2F%3E%3Cstop%20offset%3D'1'%20stop-color%3D'rgb(200%2C90%2C60)'%20stop-opacity%3D'0.03'%2F%3E%3C%2FlinearGradient%3E%3C%2Fdefs%3E%3Crect%20width%3D'320'%20height%3D'200'%20fill%3D'url(%23g)'%2F%3E%3Ccircle%20cx%3D'250'%20cy%3D'56'%20r%3D'92'%20fill%3D'rgb(200%2C90%2C60)'%20fill-opacity%3D'0.06'%2F%3E%3Ccircle%20cx%3D'64'%20cy%3D'172'%20r%3D'58'%20fill%3D'rgb(200%2C90%2C60)'%20fill-opacity%3D'0.05'%2F%3E%3C%2Fsvg%3E)
