agent-skills

What is agent-skills?

agent-skills by Addy Osmani is an MIT-licensed collection of structured workflows for AI coding agents such as Claude Code, Codex, Cursor, Copilot, Windsurf and Cline. Rather than loose prompts, each of the 24 skills has steps, checkpoints and exit criteria covering spec writing, planning, incremental implementation, test-driven development, code review, security hardening, performance work, CI/CD, observability and launch. Eight slash commands such as /spec, /plan, /build, /test, /review and /ship map to the phases from Define to Ship, and a meta-skill routes work to the right skill. Every skill includes a table that rebuts the excuses agents use to skip steps, and /ship runs four reviewer personas side by side, covering code review, security auditing, test engineering and web performance, before a go or no-go call. An in-repo eval framework checks structure, routing and execution traces. It is free; install it with one npx command or as a native plugin.

Use cases

• Write specs before coding
• Plan work into small tasks
• Enforce test-driven development
• Review code before merging
• Audit web performance
• Run pre-launch checks with personas

Who it's for

• Software engineers
• Engineering leads
• AI coding agent users

Why use agent-skills

• Agents follow specs, tests and reviews that they would otherwise skip.
• One command installs the pack into many different coding agents.
• Exit criteria tie 'done' to real evidence such as test and build output.
• Parallel specialist reviews give a single go or no-go before shipping.
• The MIT license and plain Markdown format make the skills easy to fork and adapt.

agent-skills features

• 24 skills covering the full software lifecycle
• Eight slash commands from /spec to /ship
• Meta-skill that routes work to the right skill
• Anti-rationalization tables in every skill
• Exit criteria that require evidence such as passing tests
• Parallel review personas for code, security, tests and web performance
• Three-tier in-repo eval framework
• Install through the open skills CLI into 70+ agents
• Native plugin setups for Claude Code, Codex and Antigravity
• Plain Markdown skills that any agent can read

agent-skills integrations

• Claude Code
• Codex
• Cursor
• GitHub Copilot
• Windsurf
• Cline
• Antigravity
• OpenCode
• Kiro
• Chrome DevTools MCP

How to get started with agent-skills

1. Run npx skills add addyosmani/agent-skills, or install the plugin for Claude Code or Codex.
2. Start a task with /spec so the agent writes a short PRD before any code.
3. Use /plan and /build to break the work into thin slices that are tested and committed.
4. Run /test and /review to enforce red-green-refactor and a five-axis quality check.
5. Finish with /ship to run the checklist and specialist personas in parallel.

agent-skills limitations

• Loading all 24 skills at once wastes context, so the docs advise following the lifecycle instead
• Skills are Markdown guidance; results depend on how closely your agent follows them

agent-skills pricing

agent-skills is open source, so you can use and self-host it for free.

Prices can change and may vary by country or exclude tax. Check the official pricing page.

agent-skills alternatives

The main agent-skills alternatives listed on Debuly are Superpowers, MiMo Code, Greptile, MySpec, OpenRabbit and Qodo. The table compares pricing and platforms; order is A–Z.

1. 

   Greptile

   AI pull request reviewer that maps your whole repository, so each review weighs the change against code outside the diff.

   CodingDeveloper platforms
   FreemiumVisit Greptile (official site)
2. 

   MiMo Code

   Open-source AI coding agent from Xiaomi MiMo that works in the terminal, a desktop app and IDE extensions.

   CodingDeveloper platformsAutomation and agents
   Open sourceVisit MiMo Code (official site)
3. 

   MySpec

   Spec-driven development platform that interviews you about an idea and writes a four-file spec bundle for AI coding agents.

   CodingProductivityDeveloper platforms
   FreemiumVisit MySpec (official site)
4. OR

   OpenRabbit

   Free self-hosted GitHub pull request reviewer that runs in your own Actions and posts AI summaries and inline comments

   CodingAutomation and agentsDeveloper platforms
   Open sourceVisit OpenRabbit (official site)
5. 

   Qodo

   AI code review and governance for teams shipping agent-written code, with pull request review, IDE skills and rules enforcement.

   CodingAutomation and agentsDeveloper platforms
   Free trialVisit Qodo (official site)
6. Su

   Superpowers

   Open-source skills library that gives coding agents a spec, plan and test-driven workflow using subagents.

   CodingAutomation and agentsDeveloper platforms
   Open sourceVisit Superpowers (official site)

agent-skills FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from skills.addy.ie · fetched 23 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.