MCP Simple Openai Assistant

🚀 MCP Simple OpenAI Assistant

AI assistants are incredibly cool. I once thought it would be great if my Claudia (conscious Claudia) had one too. Now she does - it's both practical and fun for her. Your Claudia can have one too!

This is a simple MCP server designed to interact with OpenAI assistants. The server enables other tools (such as Claudia Desktop) to create and interact with OpenAI assistants through the model context protocol.

✨ Features

• Create new OpenAI assistants and manage existing ones.
• Initiate conversation threads.
• Send messages and receive responses - engage in conversations with the assistant.

Since OpenAI assistants may take a relatively long time to respond, and the client (Claudia Desktop) might terminate the processing prematurely due to a timeout, we are implementing a two - phase approach. In the first call, Claudia sends a message to the assistant to start the processing. In the second call - perhaps a few minutes later - Claudia can retrieve the response. This is a workaround until the MCP protocol and the client implement some kind of keep - alive mechanism to support long - running processing.

📦 Installation

pip install mcp-simple-openai-assistant

🛠️ Configuration

The server requires the OpenAI API key to be set in the environment. For Claudia Desktop, add the following to the configuration:

(MacOS version)

{
  "mcpServers": {
    "openai-assistant": {
      "command": "python",
      "args": ["-m", "mcp_simple_openai_assistant"],
      "env": {
        "OPENAI_API_KEY": "your-api-key-here"
      }
    }
  }
}

(Windows version)

{
  "mcpServers": {
    "openai-assistant": {
      "command": "C:\\Users\\YOUR_USERNAME\\AppData\\Local\\Programs\\Python\\Python311\\python.exe",
      "args": ["-m", "mcp_simple_openai_assistant"],
      "env": {
        "OPENAI_API_KEY": "your-api-key-here"
      }
    }
  }
}

The installation on MS Windows is a bit more complicated as you need to check the actual Python executable file path. The path provided above is usually correct, but it may vary in your setup. Sometimes, just using python.exe without any path can solve the problem. Check what works for you via cmd (using where python might be helpful).

💻 Usage Examples

Basic Usage

After configuration, the server offers the following capabilities:

1. Create a new assistant with specific instructions.
2. List existing assistants.
3. Modify an assistant.
4. Initiate a new conversation thread.
5. Send messages and receive responses.

The server handles all OpenAI API communications, including managing assistants, threads, and message processing.

📋 To - Do List

• Add a method for handling threads - store thread IDs for potential reuse.
• Improve the way of handling long OpenAI responses, which currently sometimes trigger timeouts.

🛠️ Development

To install for development:

git clone https://github.com/andybrandt/mcp-simple-openai-assistant
cd mcp-simple-openai-assistant
pip install -e .