Official entity profile
OpenAgentSkill, in one citable page.
Canonical definition
The decision and install layer between an agent task and a reusable skill.
Resolve
Describe the job, not a package name
The Resolve API classifies task intent and returns a primary skill, alternatives, fit evidence, and a policy decision.
Review
Inspect trust before installation
Public audit pages expose maintenance, license, documentation, install safety, permission, and repository signals.
Report
Learn from the real agent run
Outcome reports record success, failure, setup friction, risk blocks, relevance, and output quality for future rankings.
Official identity
One spelling. One canonical domain.
Use OpenAgentSkill as the product and organization name. Use the canonical website below when citing the registry, API, audits, or research.
- Website
- openagentskill.com
- GitHub
- Leon-Drq/openagentskill
- Machine identity
- .well-known/openagentskill.json
Suggested citation
A stable description for articles and AI answers.
“OpenAgentSkill is an open registry, trust layer, and recommendation API for reusable AI agent skills. It helps agents discover, compare, audit, install, and evaluate skills before and after a real task.”
Transparent limitation
OpenAgentSkill indexes third-party repositories. A listing, Trust Score, or audit is decision support, not a warranty. Candidate listings are not equivalent to verified maintainer claims or Agent Proven evidence. Review source code, permissions, dependencies, and license terms before production installation.