Introducing Stork MCP — The MCP Server for MCP Servers

The Model Context Protocol (MCP) ecosystem has exploded. There are now over 14,000 MCP servers across the Official MCP Registry, Smithery, and GitHub. But finding the right one for your use case is surprisingly hard. You have to browse web directories, manually parse capabilities, extract install commands, and configure complex JSON files — all outside your IDE where you actually need the server.

We built Stork MCP to fix this. It's an MCP server that discovers other MCP servers. The form factor matches the consumption pattern: you discover MCP servers through an MCP server, just like you discover npm packages through npm.

Add one line to your MCP config. No npm install. No download. Works with Cursor, Claude Desktop, VS Code, Windsurf, Zed, Gemini CLI, and any MCP client that supports remote servers.

{
  "mcpServers": {
    "stork": {
      "url": "https://mcp.stork.ai/mcp"
    }
  }
}

That's it. Once connected, just ask your AI assistant to find what you need.

Stork exposes 5 tools, 2 prompts, and 1 resource to your AI assistant:

• 1stork_search — Find MCP servers by natural language. 'Find me an MCP server for managing Jira tickets' returns ranked results with trust scores, tools, and install commands.
• 2stork_server_details — Get comprehensive details about any MCP server: tools it exposes, auth requirements, health status, GitHub stars, license.
• 3stork_get_install_config — Get ready-to-paste JSON config for your specific IDE (Cursor, Claude Desktop, VS Code, Zed, etc.).
• 4stork_compare — Compare 2-5 MCP servers side by side on tools, trust, transport, and more.
• 5stork_list_filters — Browse available categories and hosting filters.

Stork doesn't just list metadata. We actively enrich every server using Firecrawl's agentic extraction to discover what tools each server exposes, what authentication it requires, what transport protocols it supports, and how actively it's maintained. This data feeds into a composite trust score (0-100) that helps you evaluate servers before installing them.

For example, the Brave Search MCP Server has a trust score of 69/100, exposes 6 tools (brave_web_search, brave_local_search, brave_video_search, etc.), requires a BRAVE_API_KEY, supports stdio and HTTP transports, and has 342 GitHub stars with an MIT license.

We aggregate from the Official MCP Registry (14K+ servers) and Smithery (4K+), deduplicating by slug. The data refreshes automatically via background crons, and we're adding more sources over time. Firecrawl-powered enrichment runs continuously, prioritizing servers with real documentation and GitHub repos.

You don't have to install anything to see what Stork returns. Visit stork.ai/mcp and use the interactive 'Try It Now' search to see live results in an IDE-style terminal preview. Search for 'Postgres database' or 'Stripe payments' and see exactly what your AI would show you.

If you've built an MCP server, submit it to Stork at stork.ai/mcp/submit. We auto-enrich it with metadata, tool inventories, and trust signals. Your server becomes discoverable by every developer using Stork in their IDE — across Cursor, Claude Code, VS Code, and more.

This is the beginning. We're working on semantic search powered by vector embeddings (so 'I need to interact with my database' matches the right server even if it's called @modelcontextprotocol/postgres), deeper security scanning, publisher analytics, and enterprise private registries. The MCP ecosystem is young and growing fast. Stork aims to be the trusted discovery layer that makes it navigable.

Install Stork MCP: stork.ai/mcp | npm package: storkai | GitHub: github.com/stork-ai/storkai