APISIX Model Context Protocol (MCP) server bridges LLM-based AI clients with the APISIX Admin API, enabling natural-language management of APISIX resources.
https://github.com/api7/apisix-mcpIf you're managing APISIX routes, plugins, and services, you know the drill: memorize the Admin API syntax, craft JSON payloads, remember endpoint paths, debug configuration errors. There's got to be a better way than constantly referencing docs just to create a simple route.
Managing APISIX typically means context-switching between your code and the Admin API documentation. Want to add rate limiting to a route? That's a specific JSON structure with exact plugin names. Need to update an upstream? Hope you remember the correct endpoint and payload format. Testing a new service route? Time to construct the perfect curl command again.
Every configuration change becomes a mini research project instead of a quick task.
The APISIX MCP Server bridges your AI assistant directly with APISIX's Admin API. Instead of crafting API calls, you just describe what you want:
Instead of this:
curl -X PUT http://127.0.0.1:9180/apisix/admin/routes/1 \
-H "X-API-KEY: your-api-key" \
-H "Content-Type: application/json" \
-d '{
"uri": "/api/users/*",
"upstream": {
"type": "roundrobin",
"nodes": {
"user-service:8080": 1
}
},
"plugins": {
"limit-req": {
"rate": 10,
"burst": 20
}
}
}'
You can say:
"Create a route for /api/users/* that proxies to user-service:8080 with rate limiting of 10 requests per second"
Quick Service Testing: "Send a test request to my new user service route" - no need to remember the gateway URL and construct test calls.
Plugin Experimentation: "Add JWT authentication to the admin routes" or "Enable CORS for the API endpoints" - your AI assistant handles the plugin configuration syntax.
Environment Synchronization: "Show me all routes in production" then "Create the same user service route in staging" - compare and sync configurations conversationally.
Debug Sessions: "Why is my payment service route returning 502?" followed by "Check the upstream health for payment-service" - troubleshoot through natural conversation.
This isn't just basic route management. The MCP server exposes APISIX's full Admin API through conversational commands:
You can manage your entire APISIX deployment through your AI assistant - from simple route creation to complex plugin configurations and security policies.
Works with Claude, Cursor, and any MCP-compatible AI client. One npx command gets you connected:
npx -y @smithery/cli install @api7/apisix-mcp --client claude
Point it at your APISIX instance with environment variables, and you're ready to manage your API gateway conversationally. The server handles all the Admin API authentication, JSON formatting, and error handling behind the scenes.
If you're running microservices with APISIX, gateway management shouldn't slow down your deployment pipeline. This MCP server turns your AI assistant into an APISIX expert, making configuration changes as quick as describing them.
Perfect for teams that want to maintain infrastructure through code review processes but need quick, reliable gateway changes during development and debugging.