Overview
What is Chatgpt?
This MCP (Model Context Protocol) stdio server forwards prompts to OpenAI’s ChatGPT (GPT-4o) for advanced summarization, analysis, and reasoning. It is designed to run inside LangGraph-based assistants.
How to use Chatgpt?
Build and run the Docker container with your OPENAI_API_KEY, or run the Python script directly using the --oneshot flag. Configure the server via an mcpServers JSON block, setting the command to python3 server.py --oneshot and providing the API key as an environment variable. The only exposed tool is ask_chatgpt, which takes a content string.
Key features of Chatgpt
- Exposes a single tool:
ask_chatgpt - Sends text to GPT-4o for external reasoning
- Supports one-shot stdin/stdout mode
- Deployable via Docker or Python directly
- API key injected securely through environment variables
Use cases of Chatgpt
- Summarize long documents
- Analyze configuration files
- Compare multiple options
- Perform advanced natural language reasoning
FAQ from Chatgpt
What tool does this server expose?
It exposes the ask_chatgpt tool, which forwards the provided content to GPT-4o for analysis or summarization.
How do I provide my OpenAI API key?
Set the OPENAI_API_KEY environment variable, either in a .env file (auto‑loaded by python‑dotenv) or by exporting it directly.
Can I test the server locally?
Yes, use the --oneshot flag and send a JSON‑formatted request via echo, as shown in the README manual test example.
How do I integrate Chatgpt with LangGraph?
Configure the server with command python3, args ["server.py", "--oneshot"], and the OPENAI_API_KEY in the environment block of your LangGraph pipeline.
What dependencies does it require?
The server depends on openai, requests, and python-dotenv; these are installed during the Docker build.
