home / mcp / ask mcp - hosted openai mcp server (v0.3.0)
Provides a hosted MCP server that connects your IDE to OpenAI for questions, plan reviews, and health checks.
Configuration
View docs{
"mcpServers": {
"bernierllc-brain-trust-mcp": {
"url": "http://localhost:8000/mcp",
"headers": {
"OPENAI_API_KEY": "YOUR_OPENAI_API_KEY"
}
}
}
}Ask MCP provides a hosted MCP server that connects your IDE to OpenAI for intelligent questions, structured plan reviews, and health checks. It enables you to ask questions, review plans with a formal framework, and verify server status all through a lightweight, per-request authentication workflow.
You connect your MCP client to the brain-trust server to start using three practical tools. Use phone_a_friend to ask OpenAI questions with optional context. Use review_plan to get AI-powered evaluation of planning documents across multiple depths using the Master Review Framework. Use health_check to verify server status and configuration. All interactions pass your OpenAI API key per request, so you don’t store keys on the server.
Prerequisites you need before you start: Python 3.12+, an OpenAI API key, and Docker (recommended for quickest setup). You will use a local Python run or a Docker deployment to start the MCP server.
# Option 1: Docker (Recommended)
# Ensure you are in your workspace with the MCP server files present
# Start the services
docker-compose up -d
# View logs
docker-compose logs -f
# Check health endpoint
curl http://localhost:8000/health# Option 2: Local Python
# Install dependencies
pip install -r requirements.txt
# Run the server
python server.py
# Server runs on http://localhost:8000The server is designed to receive your API key per request from your MCP client. You configure your client with the OPENAI_API_KEY, and the server uses that key to call OpenAI for each tool invocation. This keeps keys out of the container environment and supports per-client key management.
Ask OpenAI a question with optional context to tailor the answer.
Review planning documents using the Master Review Framework with multiple depth levels.
Check server health and configuration to ensure the MCP server is operating correctly.