Agent Skills provides production-grade engineering workflows as a collection of 24 skills for AI coding agents, with multiple installation paths and broad IDE/plugin support. It includes an open-source repository with 81,574 stars and active maintenance as of 2026-08-04.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Agent Skills offers a suite of structured workflows and anti-rationalization guidelines that encode senior engineering practices for AI coding agents. The project packages 24 skills (23 lifecycle skills plus a meta-skill) to apply guidance across define, plan, build, verify, review, and ship stages. The README describes a visual workflow and a set of slash commands that map to the development lifecycle.
How it works
Skills are organized as modular components with dedicated SKILL.md files under skills/, each providing a specific workflow (e.g., interview-me, spec-driven-development, incremental-implementation, code-review-and-quality). The repository describes auto-activation of skills based on task context (e.g., API design triggers api-and-interface-design; building UI triggers frontend-ui-engineering). There are multiple integration paths (Claude Code, Cursor, Antigravity, Gemini CLI, Windsurf, OpenCode, GitHub Copilot, Codex, Kiro), each with installation steps or usage notes.
Getting started
Quick Start instructions use the open skills CLI to install all skills:
npx skills add addyosmani/agent-skills # install all 24 skills
npx skills add addyosmani/agent-skills --list # browse before installing
Or install individual skills:
npx skills add addyosmani/agent-skills --skill code-review-and-quality
npx skills add addyosmani/agent-skills --skill interview-me
npx skills add addyosmani/agent-skills --skill test-driven-development
The README also provides per-robot/tool guidance, including Claude Code and Codex integration examples.
Recent releases
Latest releases listed with brief descriptions:
- 0.6.6 (2026-08-04): fix release restoring four agent personas for Claude Code plugin and ecosystem-neutral pass on performance.
- 0.6.5 (2026-07-26): eval framework moved to CI-enforced; validators hardened; ecosystem-neutral pass to stop assuming a JavaScript stack.
- 0.6.4 (2026-07-12): in-repo skill eval framework and native Codex support; security hardening; new skill content and adoption/contrib updates.
- 0.6.3 (2026-07-03): improved code review, new reference checklists, guardrails to reduce duplication, and related updates.
- 0.6.2 (2026-06-11): changelog includes marketplace install path fix and docs corrections.
Traction
GitHub repository shows 81,574 stars, 8,789 forks, and 1,42 open issues. The repo is JavaScript and MIT-licensed, created 2026-02-15 with last push on 2026-08-04.
Behind the repo
The project appears under the addyosmani organization and lists multiple integrations with AI agents and CLIs (Claude Code, Cursor, Antigravity, Gemini, Windsurf, Codex, Copilot, Kiro). The repository describes installation commands and cross-tool setup paths for these integrations.
Caveats
License: MIT. Created 2026-02-15. Last push 2026-08-04. Open issues: 142. Language: JavaScript. The README details multiple vendor-specific installation steps and potential SSH vs HTTPS cloning considerations for Claude Code marketplace integration.