Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
This collection provides structured guidance for AI agents to handle complex software engineering tasks. It establishes a framework for maintaining stability and quality across the development lifecycle, from initial architecture to deployment.
By implementing these skills, agent builders can standardize how their models approach API design, context engineering, and CI/CD automation, ensuring that AI-generated contributions adhere to professional engineering rigor.
WHO IT'S FOR
AI agent / automation builders
implementing production-grade engineering skills for agents
Platform / DevEx teams
standardizing AI-driven CI/CD and quality gates
Frontend / Fullstack engineers
debugging browser runtimes via AI agents
Technical leads / architects
defining stable API contracts and module boundaries
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
Install code-review-and-quality with the skills CLI
Installs: Only the code-review-and-quality skill; shared repository-level references are not copied by this per-skill installation · Claude Code, Cursor, Codex, Copilot, Cline, 70+ agents supported by the skills CLI
Run this command in a terminal. Be aware that the per-skill install copies only the selected skill folder, so supplementary shared checklists under the repository-level references directory will be unavailable unless supplied separately as described by the author.
Installs: Only the interview-me skill; shared repository-level references are not copied by this per-skill installation · Claude Code, Cursor, Codex, Copilot, Cline, 70+ agents supported by the skills CLI
Run this command in a terminal. Be aware that the per-skill install copies only the selected skill folder, so supplementary shared checklists under the repository-level references directory will be unavailable unless supplied separately as described by the author.
npx skills add addyosmani/agent-skills --skill interview-me # requirements interrogation, one question at a time
Install test-driven-development with the skills CLI
Installs: Only the test-driven-development skill; shared repository-level references are not copied by this per-skill installation · Claude Code, Cursor, Codex, Copilot, Cline, 70+ agents supported by the skills CLI
Run this command in a terminal. Be aware that the per-skill install copies only the selected skill folder, so supplementary shared checklists under the repository-level references directory will be unavailable unless supplied separately as described by the author.
Manually browse the cloned skills/ directory and choose the relevant skill's SKILL.md file.
Manually copy the selected SKILL.md content into the agent's system prompt, project rules file, or conversation. If the host already discovers skills from their descriptions, install individual skills and let the host activate them instead of adding using-agent-skills as a second always-on router.
Use the idea-refine skill to help me refine this idea: [describe the idea]. First restate it, ask sharpening questions, generate a few variations, and surface hidden assumptions. Do not save the final one-pager until I confirm the direction.
Turn an approved specification into an implementation plan
Uses planning-and-task-breakdown
Path to an approved specification
Path to the project repository
Any known delivery constraints
Use the planning-and-task-breakdown skill. Read the approved specification at [spec path] and the relevant code under [repository path], then draft an ordered set of small, verifiable tasks with dependencies and explicit acceptance criteria. Stay in read-only planning mode and do not implement the tasks.
Use the debugging-and-error-recovery skill to triage this failure: [failing command or reproduction steps]. Inspect [error output or log path], reproduce the problem if possible, preserve the relevant evidence, and report the most likely root cause plus the next verification step. Do not change code yet.
Claude
Codex
Cursor
Gemini CLI
Copilot
Windsurf
OpenCode
Kiro
Antigravity
Cline