OpenAI

This MCP server allows Claude to query OpenAI models, requiring an OpenAI API key and offering read-only access to model outputs.

Connections & Capabilities

Connects To

GitHub

Capabilities

read

Quickstart

Install

pip install -e

Config

{
  "mcpServers": {
    "openai-server": {
      "command": "python",
      "args": ["-m", "src.mcp_server_openai.server"],
      "env": {
        "PYTHONPATH": "C:/path/to/your/mcp-server-openai",
        "OPENAI_API_KEY": "your-key-here"
      }
    }
  }
}

Exposed MCP Tools (1)

safe

query_openai_model

Queries a specified OpenAI model with a given prompt and returns the response.

Read-only access to model outputs; no destructive operations.

Safety Assessment

This server is relatively safe for querying OpenAI models, but the security of the OpenAI API key is crucial. Ensure the API key is stored securely and rotated regularly. Prompt injection vulnerabilities should be mitigated through careful input validation.

Read-only access to OpenAI model outputs from Claude.
Requires an OpenAI API key for authentication.
No direct file system access or shell execution.
Relies on OpenAI's security measures for model execution.

Exposure of OpenAI API key if not properly managed.
Potential for prompt injection vulnerabilities if input sanitization is insufficient.
Reliance on OpenAI's service availability and security.
Limited control over the underlying model execution environment.