OpenAI Assistant Integration MCP Server

Developer ToolsPython

Enables Claude to create and interact with OpenAI assistants

Available Tools

create_assistant

Creates a new OpenAI assistant with specified instructions

list_assistants

Lists existing OpenAI assistants

modify_assistant

Modifies an existing OpenAI assistant

start_thread

Starts a new conversation thread with an assistant

send_message

Sends a message to an assistant and initiates processing

get_response

Retrieves the response from an assistant after processing

This MCP server bridges Claude with OpenAI's assistant capabilities, allowing Claude to create, manage, and converse with OpenAI assistants. It implements a two-stage approach to handle potentially long response times from OpenAI assistants, making it practical for real-world use. The integration provides a seamless way for Claude to leverage specialized OpenAI assistants for specific tasks, expanding Claude's capabilities through this complementary AI service. Users can create new assistants with custom instructions, manage existing ones, and conduct conversations through Claude's interface.

Overview

The OpenAI Assistant Integration MCP allows Claude to create and interact with OpenAI's assistants. This integration enables Claude to leverage specialized OpenAI assistants for specific tasks, expanding its capabilities through complementary AI services.

Installation

Option 1: Install via Smithery (Recommended)

The easiest way to install this MCP is through Smithery:

npx -y @smithery/cli install mcp-simple-openai-assistant --client claude

Option 2: Manual Installation

If you prefer to install manually:

Install the package using pip:

pip install mcp-simple-openai-assistant

Configure Claude Desktop by adding the MCP server to your configuration file.

Configuration

This MCP requires an OpenAI API key to function. You'll need to add this to your Claude Desktop configuration.

For macOS:

Add the following to your Claude Desktop configuration:

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

For Windows:

The Windows configuration requires specifying the full path to your Python executable:

{
  "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"
      }
    }
  }
}

Note: The Python path may differ on your system. You can find your Python path by running where python in Command Prompt.

Usage

Once configured, Claude can use the OpenAI Assistant Integration to:

Create new OpenAI assistants with specific instructions
List and browse existing assistants
Modify assistant configurations
Start new conversation threads
Send messages to assistants and receive their responses

The MCP handles all communication with the OpenAI API, including managing assistants, threads, and message handling.

How It Works

This MCP implements a two-stage approach to handle potentially long response times from OpenAI assistants:

In the first call, Claude sends a message to the assistant to start processing
In a subsequent call (which may be several minutes later), Claude retrieves the response

This approach works around timeout limitations in the current MCP protocol implementation.

Development

For developers who want to contribute or modify the MCP:

Clone the repository:

git clone https://github.com/andybrandt/mcp-simple-openai-assistant

Navigate to the directory:
```
cd mcp-simple-openai-assistant
```
Install in development mode:
```
pip install -e .
```