Find reusable instructions for your project, inspect their complete files, and keep an exact skill set you can review and reuse.
Agentic Awesome Skills is a library of 2,872+ installable SKILL.md playbooks. AAS Core helps Codex or Claude search the complete local catalog, record the skills the agent chooses, and preview a plan you can inspect before changing a target. Core does not rank or recommend skills.
Current release: V20.1.0. AAS Core supports local catalog inspection, agent-owned selection, stack validation, and plan preview. Apply and recovery remain experimental. Read the AAS Core preview guide for setup and exact trust boundaries.
This README tracks main. Features listed under Unreleased require a later release; the versioned guide describes the published package.
This is an independent community project, not affiliated with or endorsed by Google. Google, Antigravity, Gemini, and related names describe compatibility and install targets. The GitHub repository is canonical; the hosted catalog and browser-local Workbench are companion discovery and review surfaces.
A 40-second introduction to Agentic Awesome Skills. Press play to watch it here on GitHub.
https://github.com/user-attachments/assets/02aa20ca-c3bb-4984-807e-7b06ef77e785
If GitHub's player stalls, watch the video on the AAS website.
Help keep the catalog open, maintained, and available to the builders who use it.
| Support AAS | What it helps fund |
|---|---|
| ♥ Sponsor AAS | Ongoing maintenance, reviews, and release work |
| Buy Me a Coffee | Small, direct contributions from the community |
Every contribution helps us keep the skills curated, the tooling tested, and the catalog useful for the next project.
Security tooling support: Snyk.
AI compute & model credits: Atlas Cloud (supporting media generation workflows via atlas-cloud-media).
This project is tested with BrowserStack.
For a complete setup through your first skill selection and installation preview, follow Codex setup or Claude Code setup.
Codex or Claude inspects your project and chooses exact skills. Every current catalog skill is individually searchable, readable, and available for agent selection. The read-only compose_stack tool validates the chosen IDs and structure in memory; a client or the aas CLI can persist aas-stack.json and optional selection evidence. aas stack validate checks the manifest, while aas stack plan writes an immutable preview for your review. Manifests have a technical maximum of 128 skills; Core does not rank or recommend candidates or install skills in this supported preview path.
[!IMPORTANT] Structural and identity validity does not certify semantic fit, compatibility, setup correctness, operational safety, or safety to apply. Apply and recovery require experimental opt-in and remain outside the supported preview.
The Workbench reviews stack and plan artifacts in browser memory without accessing your filesystem. See the Core guide for tool contracts, capability coverage, and limits.
Start with AAS Core in Codex or Claude. Configure the local MCP using the Codex or Claude guide. With the MCP available, ask the agent to inspect your project, compare relevant skills, and save the exact selection. Then validate its manifest and review the resulting plan before any installation. The first configuration command previews a change and returns an approval digest:
npm exec --yes --ignore-scripts --package=agentic-awesome-skills@20.1.0 -- aas mcp configure \
--host codex \
--scope user \
--config /absolute/path/to/codex/config.toml \
--cache-root /absolute/path/to/aas-cache
Use --host claude and its configuration path for Claude. The Core setup guide explains approval, reconnection, validation, and planning. To hand the reviewed IDs to the direct installer, use aas stack install-preview as described in the manifest handoff; that command only prepares a --dry-run preview and does not apply a Core plan.
If you already know the IDs, preview a focused install into your host's skill directory:
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
Review the preview, then repeat without --dry-run when ready. The direct installer does not consume or apply a Core plan. Antigravity's watched skill directory can overload its context, so its default target requires a selected set, a filter, or an explicit --all override. See the installation guide and security guidance for other targets, auditing, and failure modes.
Use the path for your agent. Core setup is available for Codex and Claude; the other rows show direct install targets or plugin options.
| Tool | Install | First Use |
|---|---|---|
| Claude Code | AAS Core local MCP preview, direct install, or Claude plugin marketplace | Ask Claude to choose and compose an AAS stack |
| Cursor | npx agentic-awesome-skills --cursor |
@brainstorming help me plan a feature |
| Gemini CLI | npx agentic-awesome-skills --gemini |
Use brainstorming to plan a feature |
| Codex CLI | AAS Core local MCP preview or npx agentic-awesome-skills --codex |
Ask Codex to choose and compose an AAS stack |
| Autohand Code | npx agentic-awesome-skills --path ~/.autohand/skills or --path .autohand/skills |
Use brainstorming to plan a feature |
| Antigravity IDE | npx agentic-awesome-skills --antigravity --skills <ids> --dry-run |
Ask an MCP-enabled agent to choose exact IDs first |
Antigravity CLI (agy) |
`npx agentic-awesome |
| 日期 | 榜单 | 排名 | 周期新增 | 总 STAR |
|---|