Skip to content

MCP Integration

Weproxa includes an embedded Model Context Protocol (MCP) server that exposes proxy tools to external AI clients, enabling AI-assisted HTTP debugging.

MCP is an open protocol that allows AI models to interact with tools and data sources. Weproxa’s MCP server lets AI assistants access your captured HTTP traffic and proxy tools.

  1. Open Settings in Weproxa
  2. Navigate to the MCP section
  3. Enable the MCP server
  4. Note the server URL and authentication token displayed

Configure your MCP-compatible AI client with:

  • Server URL — the local HTTP endpoint (e.g., http://127.0.0.1:<port>)
  • Authentication — Bearer token (shown in Weproxa settings)

The MCP server exposes tools that let AI assistants:

  • List captured requests
  • Inspect request/response details
  • Filter requests by various criteria
  • Interact with proxy features
  • The MCP server only listens on 127.0.0.1 (localhost)
  • Bearer token authentication is required for all requests
  • The server is disabled by default