Vet is an MCP server published by ai.getvet that helps agents and developers discover MCP servers and AI tools. It connects to a catalog of over 200,000 AI tools and exposes search and vetting capabilities, including security verification and access to tool schemas. The server communicates over stdio transport and is distributed as the npm package @getvetai/mcp at version 0.4.2.
To run Vet, two environment variables must be set: VET_API_KEY for authentication and VET_API_URL to point at the API endpoint. No tool list is published in the registry record, so the exact callable operations are not confirmed beyond what the description states: searching the catalog and retrieving security and schema metadata for individual tools.
Vet is most relevant when an agent needs to programmatically look up what MCP servers exist, check whether a given tool has been security-verified, or retrieve its schema before deciding to use it. It is essentially a discovery and due-diligence layer for the broader MCP ecosystem.
Because no tool definitions are exposed in the registry record, developers should consult the repository at https://github.com/getvetai/vet for the current API surface before relying on specific callable methods in production.