Agent Skills by Addy Osmani

Agent Skills is a public skill pack from Addy Osmani (GitHub: addyosmani). The README says it packages the workflows, quality gates and best practices senior engineers use, so AI agents follow them at every phase of development. The repo holds 25 skills across six stages, define, plan, build, verify, review and ship, with nine slash commands to match: /spec writes the spec before any code, /plan breaks work into small tasks, /test proves it works, /review checks before merge, /ship takes it live. Skills also trigger on their own from what you're doing: designing an API calls api-and-interface-design, building UI calls frontend-ui-engineering. The README says one command with the skills CLI installs it into more than 70 agents, including Claude Code, Cursor, Codex and Copilot, and lists native setup for each tool. What sets it apart is that the whole development lifecycle is wired into one line, and every skill ships with steps, verification gates and a table of excuses it won't accept.

Who it's for:Developers who code with AI and want it to follow a real process. Not useful if you don't write code.

Before you start: Install an AI coding tool that supports Agent Skills, then follow the README. Docs are in English. The README notes that installing a single skill leaves out the repo-level shared reference files; the skill still works, but paths to the shared checklists won't resolve, so a whole-repo install is the fuller setup. The /build auto mode runs every task after you approve the plan once; the README says each task is still tested and committed on its own, and it pauses on failures or risky steps. The code is free under the MIT license.

Based on the project's public documentation. Not tested by AI Inspiration World.