MCP Chatgpt Responses

🚀 MCP ChatGPT Server

This MCP server allows you to access OpenAI's ChatGPT API directly from Claude Desktop.

📝 Learn why I built this project: I Built an AI That Talks to Other AIs: Demystifying the MCP Hype

✨ Features

• Call the ChatGPT API with customizable parameters.
• Enable long - term conversations between Claude and ChatGPT!
• Configure model versions, temperature, and other parameters.
• Use web search to get the latest internet information.
• Automatically manage the conversation state using OpenAI's response API.
• Use your own OpenAI API key.

📦 Installation

Install via Smithery

To automatically install the ChatGPT Server in Claude Desktop via Smithery:

npx -y @smithery/cli install @billster45/mcp-chatgpt-responses --client claude

Prerequisites

• Python 3.10 or higher.
• Claude Desktop application.
• OpenAI API key.
• uv for Python package management.

Installation Steps

1. Clone this repository:

   git clone https://github.com/billster45/mcp-chatgpt-responses.git
   cd mcp-chatgpt-responses
   

2. Set up a virtual environment and install dependencies using uv:

   uv venv
   

   uv install
   

3. After installation, run the following command to start the server:

   uv run server:app
   

📚 Configuration

To configure the MCP server in Claude Desktop, refer to the example configuration:

{
  "mcpServers": {
    "chatGpt": {
      "type": "http",
      "url": "http://localhost:5000"
    }
  },
  "tools": {
    "chatGpt": {
      "command": "$ chatgpt tool",
      "description": "Converse with ChatGPT. Use ${message} to pass the message.",
      "args": [
        {
          "name": "message",
          "type": "string",
          "description": "The message to send to ChatGPT.",
          "required": true
        }
      ]
    }
  }
}

💻 Usage Examples

Basic Usage

Basic Tool: ask_chatgpt

Format:

{
  $: "chatgpt tool",
  args: {
    message: string,
    options?: {
      model?: string,
      temperature?: number,
      maxTokens?: number,
      responseId?: string
    }
  }
}

Example:

{
  $: "chatgpt tool",
  args: {
    message: "Explain quantum computing.",
    options: {
      model: "gpt-3.5-turbo",
      temperature: 0.7,
      maxTokens: 1000,
      responseId: "response_123"
    }
  }
}

Advanced Usage

Tool with Web Search: ask_chatgpt_with_search

Format:

{
  $: "chatgpt tool",
  args: {
    message: string,
    options?: {
      model?: string,
      temperature?: number,
      maxTokens?: number,
      responseId?: string
    }
  }
}

Example:

{
  $: "chatgpt tool",
  args: {
    message: "What's the weather like in New York tomorrow? Then suggest suitable food and park activities.",
    options: {
      model: "gpt-3.5-turbo",
      temperature: 0.7,
      maxTokens: 1500
    }
  }
}

🔧 Technical Details

This tool utilizes OpenAI's response API to automatically manage the conversation state on OpenAI servers. This approach:

1. Simplifies the code by letting OpenAI handle the context.
2. Provides more reliable context tracking.
3. Improves the user experience by maintaining the message history.
4. Allows access to the latest information through the web search tool.

📄 License

MIT License