MCP client landscape

The reverse of the catalog: which MCP servers each client is configured with, ranked by the distinct public repositories observed wiring them up through that client. Derived from public configs today; ax-ray submissions add live on-machine usage.

Which servers run in which client

top servers × busiest clients · distinct repos

Browse by client

MCP configurations attributed to this client.

663 servers · 781 repos

  1. 1npm@upstash/context7-mcp99 repos
  2. 2pypimcp-server-fetch98 repos
  3. 3pypiawslabs.aws-documentation-mcp-server65 repos
  4. 4npm@modelcontextprotocol/server-sequential-thinking51 repos
  5. 5npm@playwright/mcp41 repos
  6. 6npmmcp-remote40 repos

VS Code (and Copilot) MCP configuration.

117 servers · 99 repos

  1. 1npmchrome-devtools-mcp11 repos
  2. 2remotehttps://api.githubcopilot.com10 repos
  3. 3npm@playwright/mcp8 repos
  4. 4remotehttps://mcp.context7.com5 repos
  5. 5npmshadcn4 repos
  6. 6npm@upstash/context7-mcp4 repos

Unknown / other

All 4,066

Configs we couldn't attribute to a known client.

4,066 servers · 2,748 repos

  1. 1npm@modelcontextprotocol/server-filesystem322 repos
  2. 2npm@upstash/context7-mcp312 repos
  3. 3npm@modelcontextprotocol/server-sequential-thinking242 repos
  4. 4npm@modelcontextprotocol/server-github230 repos
  5. 5npm@playwright/mcp218 repos
  6. 6npm@modelcontextprotocol/server-memory189 repos