Radar check
What we found when we opened it. Every bar is one fact, with its date.
Last code pushed 27 Oct 2026, the same day before this check.
Latest release v20.1.0 on 27 Oct 2026.
Public since January 2026, 9 months.
48,622 stars and 7,061 forks.
962 commits and 25 new issues in the last 90 days.
468 contributors. 945 pull requests merged in 90 days.
A security policy says how to report a problem. No published security advisories.
How the bars are scored
- Alive. Days since the last code push: a week or less 100, a month 90, three months 70, six months 50, a year 30, eighteen months 15, older 0.
- Releases. Age of the latest GitHub release, on a similar scale; no bar when the project publishes no releases.
- Maturity. Age of the project: eight years 100, five 90, three 70, two 55, one 40, six months 25, younger 10.
- Adoption. Stars on a log scale: 100,000 stars is 100, 10,000 is 80, 1,000 is 60, 100 is 40.
- Activity. Commits in the last 90 days: 300 or more 100, 100 85, 30 65, 10 45, any 25.
- Community. People who have contributed: 100 or more 100, 30 80, 10 60, 5 40, 2 25, one 10.
- Security. A published security policy 80, none 40; published advisories are listed, not scored.
Checked by Repo Radar on 28 Oct 2026 from the repository on GitHub.
In simple English
When you use an artificial intelligence helper to write code, it often forgets the best rules for your project. This tool gives your helper thousands of ready-made instruction files called playbooks. Your helper searches these instructions and follows them to plan your project safely. You can inspect every file before it runs on your computer. The project costs nothing to download and includes a command-line tool and local connection bridge.
What it is
Agentic Awesome Skills is a collection of nearly three thousand installable playbooks and a local control plane for AI coding assistants. It helps tools like Claude and Codex search instructions, validate project setups, and preview execution plans.
What it is good for
Solo builders who want their AI coding assistants to follow strict, professional rules for debugging, testing, and writing code without making up broken patterns.
Summary (Key Points)from the README
Key features
- Searchable catalog: Every current catalog skill can be searched, read in full, and selected by the agent. The library holds 2,872+ `SKILL.md` playbooks.
- Agent-owned selection: Codex or Claude chooses the exact skills. Core does not rank or recommend candidates.
- Read-only stack validation: The `compose_stack` tool checks chosen IDs and structure in memory, and `aas stack validate` checks the saved manifest.
- Reviewable plan preview: `aas stack plan` writes an immutable preview for inspection before any change to a target.
- Manifest size limit: Stack manifests have a technical maximum of 128 skills.
- Specialized plugins: Thirteen focused plugins, each with 8 to 10 skills (9 for some), cover areas such as web apps, security, data analytics, QA, DevOps, and API platforms. They are available for Claude Code and Codex.
- Direct installer with preview: `--dry-run` previews a focused install of named skills into a host's skill directory before anything is written.
- Broad host support: The README lists install paths for Claude Code, Cursor, Gemini CLI, Codex CLI, Autohand Code, Antigravity, Kiro, GitHub Copilot, OpenCode, AdaL CLI, and custom paths.
- Machine-readable discovery: `skills_index.json` and a v1 schema support hosts that load individual skill files.
Ways to install
| Platform or method | What it needs | Effort |
|---|---|---|
| AAS Core local MCP (Codex or Claude) | Node via npm exec, an absolute config path, and a cache root; the first command returns an approval digest | Highest: setup, approval, reconnection, then validation and planning |
| Direct installer (`npx agentic-awesome-skills`) | Node/npm and a target skill directory; `--dry-run` previews first | Low to moderate |
| Host shortcut flags (Cursor, Gemini CLI, Codex CLI, Antigravity CLI, Kiro CLI) | Node/npm and the matching flag | Low |
| Claude plugin marketplace | Claude Code | Low |
| GitHub Copilot via `gh skill install` | GitHub CLI; the README marks this as a preview | Low, but still preview |
Important
- Validity is not safety: Passing structural and identity checks does not certify semantic fit, compatibility, setup correctness, operational safety, or safety to apply.
- Apply and recovery are experimental: Both need explicit opt-in and sit outside the supported preview.
- Core and the direct installer are separate: The direct installer does not consume or apply a Core plan. The `install-preview` handoff only prepares a `--dry-run`.
- Some skills are paid or hosted: Several Beatra skills are paid, hosted services. Others spend prepaid credits, place real phone calls, or require API keys or OAuth. Some are governed by separate hosted-service terms.
- Adult content: The `nsfw-ai-spicyapi` skill covers adult (18+) generation and is only for adults with consent.
- Antigravity overload: Its watched skill directory can overload context, so the default target needs a selected set, a filter, or an explicit `--all` override.
- Mixed licenses: Entries carry different licenses, including MIT, MIT-0, Apache-2.0, AGPL-3.0, BSD-3-Clause, and FSL-1.1-ALv2.
- Preview of `main`: The README tracks `main`. Features under Unreleased need a later release, and the versioned guide describes the published package.
- Independent project: It is a community project, not affiliated with or endorsed by Google.
Written by an AI model from the project’s README; not part of the review, and it can be wrong.
- Provides thousands of specialized instruction files for many programming languages and tasks.
- Lets your AI assistant inspect your project and choose the right rules automatically.
- Includes a local preview mode so you can check every instruction before applying it.
- Requires comfort with command-line tools and Node.js package execution.
- Managing too many instruction files can overwhelm your AI helper's memory limit.
- Some specialized skills point to paid external services or require separate API keys.
Manually writing custom prompt instructions for every new coding project.
Claude Code
- anthropics/skills184k stars
Official Anthropic collection of document and workflow skills for Claude assistants.
- vercel-labs/agent-skills33k stars
Vercel Labs collection of React best practices and web design guidelines for AI agents.
- vllm-project/vllm96k stars
A serving engine that makes one graphics card answer many requests at once.
- infiniflow/ragflow93k stars
A question-answering system built on a document collection, with the sources shown beside every answer.
- basicmachines-co/basic-memory4.3k stars
Gives an AI assistant a memory that is stored as plain text files you can read.
npm exec --yes --ignore-scripts --package=agentic-awesome-skills@20.1.0 -- agentic-awesome-skills --release 20.1.0 --path .agents/skills --skills brainstorming,systematic-debugging --dry-run