Eduframe MCP Server

eduframe-mcp

Eduframe MCP server — exposes Eduframe resources as MCP tools.

Usage in VSCode

Add the following configuration to your .vscode/mcp.json file to use this MCP server with GitHub Copilot in agent mode.

Using `npx` (from GitHub)

{
  "servers": {
    "eduframe": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "github:martijnpieters/eduframe-mcp"],
      "env": {
        "EDUFRAME_API_TOKEN": "${input:eduframe_api_token}"
      }
    }
  },
  "inputs": [
    {
      "id": "eduframe_api_token",
      "type": "promptString",
      "description": "Eduframe API token",
      "password": true
    }
  ]
}

Development (from this repository)

First, build the project:

pnpm install && pnpm build

Then add the following to .vscode/mcp.json in this repository:

{
  "servers": {
    "eduframe": {
      "type": "stdio",
      "command": "node",
      "args": ["${workspaceFolder}/dist/index.js"],
      "env": {
        "EDUFRAME_API_TOKEN": "${input:eduframe_api_token}"
      }
    }
  },
  "inputs": [
    {
      "id": "eduframe_api_token",
      "type": "promptString",
      "description": "Eduframe API token",
      "password": true
    }
  ]
}

Configuration

The server requires one environment variable:

Debugging

When running the server locally, each successful tool call writes the raw API response to .last-response.json in the project root. This file can be inspected to see the complete JSON payload returned by the Eduframe API, independent of the formatted text output sent back to the MCP client.

The file is overwritten on every tool call and contains the tool name, the request parameters, and the full response:

{
  "timestamp": "2024-01-01T00:00:00.000Z",
  "tool": "get_lead",
  "request": { "id": 1 },
  "response": { "id": 1, "email": "lead@example.com", ... }
}