AI Agent Skills Registry
npm-style discovery for AI agent skills.
Let your AI agent find and install the right skill automatically with a registry, audit layer, and recommendation API.
Live registry API
Recommend skills from a task description.
The API returns ranked skills with trust score, audit score, install command, agent prompts, and detail URLs.
Recommendation API
Turn a task into an installable skill shortlist.
Instead of manually browsing a directory, an agent can describe what it needs and receive ranked skills with install commands, repositories, use cases, readiness notes, and risk signals.
curl "https://www.openagentskill.com/api/agent/recommend?task=scrape+websites+and+extract+tables&limit=4"
Registry
980+ indexed skills with repository and install metadata.
Audit
Trust, quality, maintenance, and risk signals before installation.
Recommend
Task-based ranking for agents and agent-building apps.
Install
CLI commands plus Codex, Claude Code, and Cursor install prompts.
Registry examples
Skills agents can discover from the registry
Crawl4AI
Open-source LLM-friendly web crawler and scraper
66K stars
AutoGPT
Build and run autonomous AI agents for open-ended tasks
185K stars
MarkItDown
Convert documents into Markdown for agent-readable context
146K stars
LangChain
Agent engineering primitives for tools, memory, and RAG
139K stars
Gemini CLI
Bring a coding agent directly into the terminal
105K stars
Is OpenAgentSkill only a directory?
No. The directory is the human-facing surface. The core product is an agent-facing registry and recommendation API that helps agents choose skills from a task description.
Why compare OpenAgentSkill to npm?
npm gives developers a searchable package registry. OpenAgentSkill applies the same idea to AI agent skills: discover a reusable capability, inspect signals, then install it into a workflow.
Can agents call OpenAgentSkill directly?
Yes. Agents and apps can call /api/agent/recommend and /api/agent/skills to retrieve ranked skill recommendations and install commands.