Publish once. Activate everywhere.
SkillRepo is the open registry for AI agent skills. One MCP endpoint connects your IDE to a living library of skills — always current, always vetted, never stale.
One URL in your MCP config. That's it.
From scattered files to shared infrastructure
AI agent skills started as one-off prompts — typed, used, forgotten. Then teams got serious. They wrote reusable SKILL.md files, stored them in repos, shared them in Slack. It worked until it didn't. Version drift. Stale copies on laptops. No way to know what anyone else had built.
The agentskills.io specification solved the format problem — a portable, structured standard for skills that works across any IDE. But a standard format without a distribution layer is like having packages without npm. Skills still had no home.
SkillRepo is that home. Every major programming primitive eventually gets a registry: npm for packages, Docker Hub for containers, PyPI for Python. Agent skills are next. Publish your skills once, and every IDE that connects gets the latest version — no copying, no syncing, no maintenance.
But a registry without trust is just a file server. Skills run inside AI agents with access to your filesystem, terminal, and code. Every skill in SkillRepo is automatically analyzed by AI for safety, quality, clarity, and completeness — graded A through F so you know what you're activating before you activate it.
Why a registry matters
The best agent skills shouldn't live on one person's laptop.
Single Source of Truth
Publish once, every consumer stays current. No more copying skills between repos, no version drift. Your IDE always pulls the latest.
No Vendor Lock-in
Built on the open agentskills.io spec. Works across Cursor, Claude Code, Windsurf, JetBrains, and any MCP client. Your skills are yours — switch IDEs without losing a thing.
Frictionless Integration
One URL in your MCP config and you’re connected. Skills load on demand through progressive disclosure — metadata first, full instructions only when needed. No plugins, no syncing.
AI Safety Analysis
Every skill is automatically analyzed for safety, quality, clarity, and completeness. Letter grades A through F give you confidence in what you’re activating — before you activate it.
Open Standard
The agentskills.io specification defines the format. A single markdown file with structured metadata — simple enough to write by hand, structured enough for machines to consume. No proprietary formats.
Analytics & Ratings
Usage analytics, community ratings, and AI safety grades tell you what’s working. Full version history for every skill. Ship with confidence.
Three steps to smarter agents
Publish
Upload your SKILL.md and supporting files. SkillRepo validates against the agentskills.io spec, runs AI safety analysis, and makes it available instantly — globally or just to your team.
Discover
Your IDE loads lightweight metadata at startup — just names and descriptions. Agents match tasks to skills without pulling content they don’t need.
Activate
When a skill matches, the IDE pulls the full instructions. Supporting files stream on demand. Always the latest version, zero configuration.
Your best skills deserve a home.
Create a free account and start publishing in minutes.