addyosmani /
agent-skills
A pack of 25 Markdown skills that guide AI coding agents from spec to ship
- 109.7k stars
- 9.3k stars since 3 Oct
- JavaScript
- Updated 9 days ago
- Some experience
- Agent skills & plugins
AI coding agents tend to jump straight to code, skip tests, and drift from what you asked for. This repo packages the habits of careful engineers as skills: Markdown files with steps, verification gates, and tables that counter an agent's excuses for cutting corners.
The skills follow six phases: define, plan, build, verify, review, ship. Nine slash commands such as /spec, /plan, /build, /test, /review and /ship map to those phases and switch on the right skills. Skills can also trigger on their own. Designing an API, for example, brings in the api-and-interface-design skill.
Because the skills are plain Markdown, they are not tied to one tool. The README gives install routes for Claude Code, Cursor, Codex, Gemini CLI, Antigravity, Copilot, Windsurf, OpenCode, Kiro and Command Code. It also points to the open skills CLI, which installs into many agents at once. The README names no direct alternatives.
What it does
Lifecycle slash commands
Nine commands cover spec, plan, build, test, constraints, review, web performance, code simplification and ship. Each one activates the matching skills.
Auto mode for building
/build auto writes the plan and implements every task after you approve the plan once. Each task is still test-driven and committed on its own, and it pauses on failures or risky steps.
25 structured skills
Skills cover requirements interviews, spec-driven and test-driven development, context engineering, debugging, browser testing with DevTools, and five-axis code review. Each has steps and verification gates.
Works across many agents
One npx command installs into 70+ agents, with native routes for Claude Code and Codex. Others, like Cursor and Gemini CLI, use their own setup guides.
Adoption guide
A guide covers rolling the pack out on a new project from day one, or step by step on an existing codebase.
What you could build
- Write a spec and task plan before letting an agent touch your code.
- Have an agent build a feature in small test-driven slices, each committed separately.
- Run a structured review of a change before merging it.
- Write down a project's quality bar in a CONSTRAINTS.md file and have agents stick to it.
- Debug failing tests with a fixed routine of reproduce, localize, reduce, fix and guard.
Get started
Run the npx command to install all skills into your agent, or use --list first to browse them. Installing a single skill leaves out the shared references folder, so use a whole-repo install if you want the supplementary checklists. Then try /spec on your next feature.
npx skills add addyosmani/agent-skillsnpx skills add addyosmani/agent-skills --list/plugin marketplace add addyosmani/agent-skillsCommands are copied from the project's README. Read it on GitHub for the full setup.
More in Agent skills & plugins
superpowers
A set of skills that makes coding agents plan, test and review before they write code
mattpocock/skills
Small, editable agent skills for planning, testing and reviewing code with AI
ECC
Agents, skills, hooks and memory that add an engineering workflow to coding agents
andrej-karpathy-skills
A CLAUDE.md file that tells Claude Code to ask first, keep code small, edit narrowly
anthropics/skills
Anthropic's example Agent Skills: folders of instructions Claude loads per task
ponytail
A skill that makes AI coding agents write the least code that works
gstack
Slash-command skills that give Claude Code roles like reviewer, QA and release engineer
taste-skill
Agent skills that steer AI coding tools toward better layout, type and motion
archify
Agent skill that turns a description or a codebase into an interactive HTML diagram
impeccable
A design skill for AI coding agents that steers them away from generic frontend looks