AI agent skills
Agent native
Humans need browseable pages. Agents need structured metadata, rankings, trust signals, and install handoffs. OpenAgentSkill provides both surfaces so a workflow can move from task intent to skill selection.
Open the agent entryCoverage
The catalog includes developer automation, web extraction, RAG, data analysis, security, DevOps, commerce, finance, quant research, documents, and other reusable workflows.
Browse use casesExplore
Cluster
Find crawler, extractor, browser automation, and HTML-to-markdown skills for public web data.
OpenCluster
Route PR review, CI debugging, issue triage, tests, and repository workflows to reusable skills.
OpenCluster
Compare market data, filings, portfolio, backtesting, and stock research workflows.
OpenCluster
Use football data, xG, team comparison, prediction research, and dashboard workflows.
OpenCluster
Build retrieval, document ingestion, citation, embedding, and knowledge-base workflows.
OpenCluster
Parse PDFs, extract tables, run OCR, and convert documents into agent-ready markdown.
OpenCluster
Operate web apps through Playwright-style flows, screenshots, forms, and UI verification.
OpenCluster
Scan repositories, summarize risks, review dependencies, and prepare remediation steps.
OpenQuestions
An AI agent skill is a reusable package of instructions, scripts, metadata, examples, or workflow logic that helps an agent complete a task.
GitHub search finds repositories. OpenAgentSkill turns repositories into task-oriented skill choices with quality, trust, freshness, and install-readiness signals.
No. The registry is designed for multiple agent surfaces including Codex, Claude Code, Cursor, and other runtimes.