Model Context Protocol moderate risk

OpenAI

MCP server enabling Claude to use OpenAI models via API calls, supporting multiple models and basic error handling.

Connections & Capabilities

Connects To

GitHub

Capabilities

readwriteexec

Quickstart

Config

{
  "mcpServers": {
    "mcp-openai": {
      "command": "npx",
      "args": ["-y", "@mzxrai/mcp-openai@latest"],
      "env": {
        "OPENAI_API_KEY": "your-api-key-here (get one from https://platform.openai.com/api-keys)"
      }
    }
  }
}

Exposed MCP Tools (1)

moderate

openai_chat

Sends messages to OpenAI's chat completion API and returns the response.

Allows sending arbitrary messages to OpenAI, potentially triggering unintended actions or exposing sensitive data.

Safety Assessment

This server is relatively safe for read operations, but the ability to send messages to OpenAI's API introduces moderate risk. The primary risk stems from potential API key exposure and prompt injection vulnerabilities. It's safer when used in controlled environments with trusted users.

Requires an OpenAI API key, limiting unauthorized access.
Uses a defined message passing interface.
No direct filesystem access.
Limited toolset with specific functions.

Relies on external OpenAI API, subject to its vulnerabilities.
API key is stored in environment variables, posing a risk if exposed.
Error handling is basic, potentially leading to unhandled exceptions.
No explicit input sanitization documented, raising injection concerns.
Potential for prompt injection via user-provided messages to OpenAI.