Apifox

Log in to Apifox, go to your account settings, find the API Access section, and generate a personal access token. Copy this token — you will use it as APIFOX_ACCESS_TOKEN.

Open the Apifox project you want the AI to access. The project ID is visible in the project settings page or in the URL of the project dashboard.

Edit your MCP client config (e.g. ~/.cursor/mcp.json or claude_desktop_config.json) to add the Apifox server with your access token and project ID.

{
  "mcpServers": {
    "apifox": {
      "command": "npx",
      "args": ["-y", "apifox-mcp-server@latest", "--project-id=YOUR_PROJECT_ID"],
      "env": {
        "APIFOX_ACCESS_TOKEN": "YOUR_ACCESS_TOKEN"
      }
    }
  }
}

Restart Claude Desktop, Cursor, or your chosen client so it registers the Apifox server and starts caching the API documentation locally.

If your team uses Swagger or OpenAPI Specification files instead of Apifox projects, you can point the server at an OAS URL or local file path using the --oas flag.

{
  "mcpServers": {
    "apifox": {
      "command": "npx",
      "args": ["-y", "apifox-mcp-server@latest", "--oas=https://your-api-host/openapi.json"],
      "env": {
        "APIFOX_ACCESS_TOKEN": "YOUR_ACCESS_TOKEN"
      }
    }
  }
}

To expose multiple Apifox projects to your AI assistant, add a separate mcpServers entry for each project with a unique name and the corresponding project ID.