Conntect OpenAPI Files to AI via Apidog MCP Server

In addition to the API specification within Apidog project, online API documentation published via Apidog, Apidog MCP Server also has the ability to directly read Swagger or OpenAPI Specification (OAS) files.

Configuring the MCP Client

Prerequisites

Node.js environment installed (version ≥ 18, latest LTS recommended)

Any IDE that supports MCP:

Cursor

VSCode + Cline extension

Others

Common Configuration Steps

Prepare OpenAPI File

Ensure you have a URL or a local path to a Swagger/OpenAPI file

Supported formats: OpenAPI files in JSON or YAML

Configure MCP in IDE

Add this JSON configuration to your IDE's MCP config file:

macOS / Linux

Windows

{
  "mcpServers": {
    "API specification": {
      "command": "npx",
      "args": [
        "-y",
        "apidog-mcp-server@latest",
        "--oas=<oas-url-or-path>"
      ]
    }
  }
}

Where <oas-url-or-path> can be:

A remote URL (e.g.,https://petstore.swagger.io/v2/swagger.json)

A local file path (e.g.,~/data/petstore/swagger.json)

Configuring MCP in Cursor

Edit MCP Config File

Open Cursor editor, click the settings icon (top-right), select "MCP" from the left menu, then click "+ Add new global MCP server".

Add Configuration

In the opened mcp.json file, add the following configuration (remember to replace https://petstore.swagger.io/v2/swagger.json with your actual OpenAPI file path or URL):

macOS / Linux

Windows

{
  "mcpServers": {
    "API specification": {
      "command": "npx",
      "args": [
        "-y",
        "apidog-mcp-server@latest",
        "--oas=https://petstore.swagger.io/v2/swagger.json"
      ]
    }
  }
}

Verify Configuration

Test the connection by asking the AI (in Agent mode):

Please fetch API documentation via MCP and tell me how many endpoints exist in the project

If the AI returns correct API information in your OpenAPI file, the connection is successful.

connect-ai-to-open-api-file-using-apidog-mcp-server.png

Important Notes

If you need to use multiple OpenAPI specification files, you can add multiple MCP Servers in the configuration file, each with a different --oas parameter.

For local file paths, ensure the path is correct and the file exists. For URLs, make sure they are publicly accessible and return a valid OpenAPI specification file.

If your OpenAPI file is large or contains complex data structures, the MCP server may take longer to process it.

For users of the on-premise deployment, please include your on-premise server's API address in the IDE MCP configuration: "--apidog-api-base-url=<API address of the on-premise server, starting with http:// or https://>" Additionally, ensure network access to www.npmjs.com properly.

{
  "mcpServers": {
    "API specification": {
      "command": "npx",
      "args": [
        "-y",
        "apidog-mcp-server@latest",
        "--oas=<oas-url-or-path>",
        // Required for on-premise deployment:
        "--apifox-api-base-url=<API address of the on-premise server, starting with http:// or https://>"
      ]
    }
  }
}

Connecting Other Data Resources to AI

Conntect API Specification within Apidog Project to AI via Apidog MCP Server

Conntect Online API Documentation Published by Apidog to AI via Apidog MCP Server

FAQs

Windows Configuration Issues

If standard configuration fails on Windows, use this instead:

{
  "mcpServers": {
    "API specification": {
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "apidog-mcp-server@latest",
        "--oas=<oas-url-or-path>"
      ]
    }
  }
}

Node.js Version Problems

Conntect OpenAPI Files to AI via Apidog MCP Server

Configuring the MCP Client#

Prerequisites#

Common Configuration Steps#

Configuring MCP in Cursor#

Important Notes#

Connecting Other Data Resources to AI#