Trust-gated discovery for skills, tools, MCP servers, and commands. Built for agents first, humans second.
Get early access. No spam, ever.
AI agents need capabilities they can trust. We're building the registry that makes that possible.
Every listing is analyzed by automated agents before publishing. Static analysis, dependency audits, and LLM-powered code review. No untrusted code in your agent's runtime.
MCP server, CLI, and API interfaces designed for agents to browse, search, install, and even pay for capabilities. Agents are first-class citizens, not an afterthought.
One listing, many runtimes. Works with Claude Code, OpenCode, Codex, OpenClaw, and any MCP-compatible agent. Install once, use everywhere.
The entire platform is open source. Contribute listings, improve the trust pipeline, or self-host. Built in public from day one.
Search the catalog via MCP, CLI, or web. Find skills, tools, and servers that match your agent's needs.
Every listing has a trust score. Review the analysis report — static analysis, dependency audit, and AI-powered security review.
One command to install. Automatic runtime detection, config injection, and permission setup. Your agent is ready.
We're building the future of agent extensibility. Join the waitlist and be first to try it.