Time to complete: 5 minutes
What you’ll get: A custom MCP server from your OpenAPI spec + MCP client integration ready to use
Turn any OpenAPI/Swagger specification into a fully functional MCP server with AI-optimized tools. Perfect for connecting private APIs and custom services to your MCP clients.

Quick OpenAPI-to-MCP Process

1

Upload OpenAPI Specification

Go to the Connectors page in the xpander.ai console.
  1. Click ”+ Create new connector”
  2. Upload your OpenAPI/Swagger specification file (JSON or YAML)
  3. Provide a name and description for your connector
  4. Don’t forget the server URL, especially if your original spec doesn’t have one.
Your OpenAPI spec gets automatically analyzed by the connector generator agent!
2

Configure Authentication

Set up authentication for your API:For API Key authentication:
  1. Select “API Key” as the authentication method
  2. Specify the key location (header, query parameter, etc.)
  3. Enter your API key or leave it for later configuration
Other supported methods:
  • OAuth 2.0
  • Bearer tokens
  • Basic authentication
  • No authentication
3

Review Generated Tools

The connector generator analyzes your API and creates AI-optimized tools:
  • Function descriptions optimized for LLM understanding
  • Parameter validation based on your schema
  • Response formatting for better AI consumption
  • Error handling built-in
4

Create MCP Server

Navigate to MCP Servers section in the console:
  1. Click “New MCP Server”
  2. Give your MCP server a descriptive name
  3. Go to the Tools tab, and click + Add tools
  4. Select tools from your newly generated connector
  5. Click Update changes
  6. Go to the Connect tab to get your MCP Server URL, and one-line commands to hook up your MCP server to your favorite MCP client.
5

Connect MCP Clients

Add your MCP server to your favorite MCP client using the connection details provided:
MCP Server Connect
For other MCP clients: Use the provided URL and authentication details.

What You Just Built

🎉 A complete MCP integration with:

AI-Optimized Tools

Functions described in LLM-friendly language for better tool selection

Automatic Authentication

API keys and auth handled transparently in the background

Schema Validation

Request/response validation based your OpenAPI specification

Instant Availability

Ready to use in Claude Desktop and other MCP clients immediately

Example: Connecting a Weather API

Next Step: Self-Deploy Your MCP Server

Your MCP server runs on xpander.ai’s cloud, but you can also deploy it to your own infrastructure:
  • Self-deploy to K8s Cluster - Run your MCP servers on your own Kubernetes infrastructure
  • Full control over data and scaling
  • Enterprise security with your own network policies

Troubleshooting