Skills

Jad Madi's agent skills — for Claude Code, Cursor, Windsurf, Codex, Devin, and any other agent that supports the Agent Skills format. Built from real engineering work, refined against real usage, published as I use them.

Quickstart

npx skills add jadmadi/skills

Pick the skills you want and which agents to install them on. Skills are used automatically once installed — the agent reaches for them when a task matches.

Skills

Organized into buckets under skills/. All skills below are model-invoked (the agent can reach for them on its own, based on the task, not just by typed name).

Engineering

Debugging, pre-ship checklists, and codebase documentation.

• adoo — Structured problem-solving via the ADOO method (Assess, Deconstruct, Organize, Optimize) for bugs, refactors, and incidents where jumping straight to a fix is risky.
• pre-flight-check — Pre-commit / pre-deploy checklist for backend and full-stack changes: typecheck/lint gates, API response consistency, route ordering, schema checks, and result-checking. Ships as an adaptable template.
• codebase-artifact — Scan a codebase and produce a single self-contained HTML file documenting its architecture, user journey, component hierarchy, and key code patterns.

Cloudflare

Audits for Cloudflare Workers CPU usage and D1 database performance/billing.

• cf-cpu-audit — Scan a Cloudflare Workers repo for operations that excessively consume CPU time: global scope misses, memory buffering, pure-JS crypto, blocking iterations, missing waitUntil() deferral.
• cf-d1-audit — Scan a Cloudflare Workers repo for D1 patterns that inflate billing, risk timeouts, or cause consistency bugs: missing indexes, N+1 queries, unbatched round-trips, wildcard SELECTs, and more.

For contributors

# Scaffold a new skill from the template
scripts/create-skill.sh <bucket> <skill-name>

# Quick local sanity check (frontmatter, name/description rules, body length)
scripts/validate-skill.sh <bucket>/<skill-name>

# List every skill in the repo
scripts/list-skills.sh

# Symlink every skill into this machine's agent config directories
# (~/.claude/skills, ~/.config/devin/skills, ~/.agents/skills, etc.)
scripts/link-skills.sh

See AGENTS.md for the full conventions (naming, description hygiene, cross-agent portability, when to add a new bucket) and .out-of-scope/ for additions that were deliberately rejected, and why.

Skill structure

Each skill is a directory under skills/<bucket>/<name>/ containing:

• SKILL.md — instructions for the agent (required)
• scripts/ — helper scripts for automation (optional)
• references/ — supporting documentation loaded on demand (optional)
• assets/ — templates or files used in output (optional)

Credits

This repo's conventions were learned from others, not invented here:

• mattpocock/skills (Matt Pocock) — the bucket layout (skills/<bucket>/<name>/), per-bucket README.mds, scripts/link-skills.sh / scripts/list-skills.sh, and the .out-of-scope/ convention for recording rejected ideas instead of only in commit messages are all adapted directly from his repo. His writing-great-skills skill and its glossary also shaped how the skills here are written — description hygiene, cutting duplication, and the vocabulary for what makes a skill predictable rather than just plausible-sounding.
• vercel-labs/agent-skills — the skills.sh.json grouping format and the general shape of this README (badge, "what it does / use when" summaries per skill) come from theirs.
• Agent Skills specification — the open format every skill here implements, and the source for the authoring best-practices this repo tries to follow.

If you're setting up your own skills repo, all three are worth reading directly.

License

MIT

Author

Jad Madi — github.com/jadmadi