home / mcp / mcf mcp server

MCF MCP Server

Based on https://github.com/openai/openai-apps-sdk-examples

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "gabrielchua-mcf-mcp": {
      "command": "uvicorn",
      "args": [
        "mycareersfuture_server_python.main:app",
        "--port",
        "8000"
      ]
    }
  }
}

You will run a Python-based MCP server that forwards structured tool requests to the MyCareersFuture API and returns both structured job data and widget metadata for interactive rendering when using compatible front ends.

How to use

To use this MCP server with a client that supports the MCP protocol, start the server and connect your MCP client to it. The server provides a single tool named mycf-job-list that you can call to retrieve current job listings from MyCareersFuture in a structured format. When your client receives results, you can render an interactive widget carousel if Widgets are supported by your client.

Run the server and then send a request to the mycf-job-list tool to obtain job data and widget metadata. The response includes a list of jobs with titles, employers, salary hints, and associated metadata, plus a pointer to a static widget bundle for rendering a horizontal carousel in compatible apps.

How to install

Prerequisites you need before installing and running the MCP server:

Node.js 18+

pnpm (recommended)
npm
yarn

Python 3.10+

Steps to set up and run the MCP server locally:

# 1) Set up a Python virtual environment
python -m venv .venv
source .venv/bin/activate

# 2) Install Python dependencies for the MCP server
pip install -r mycareersfuture_server_python/requirements.txt

# 3) Start the MCP server (FastAPI + uvicorn)
uvicorn mycareersfuture_server_python.main:app --port 8000

Additional notes

For widget assets, the project builds static bundles that power the MyCareersFuture widget. If you need to preview locally, you can build and serve the widget assets first, then run the MCP server to serve the tool. The server’s HTTP interface and the single tool are designed to be minimal and easy to extend.

Security and production readiness

Plan to Harden your deployment before exposing it to users. Consider adding authentication, request validation, caching, rate limiting, and proper error handling. Ensure your environment variables and API keys (if any) are secured and not exposed in client-side code.

Examples and testing

To test in a local environment, start the server on port 8000 and point your MCP client at http://localhost:8000. Call the mycf-job-list tool to retrieve current job openings from MyCareersFuture. If you are rendering widgets, ensure the client can resolve the _meta.openai/outputTemplate pointer to the bundled widget assets.

Available tools

mycf-job-list

Queries live MyCareersFuture job listings and returns structured results with widget metadata.