OpenAI FAQs

Question 1

What is the OpenAI MCP Server?

Accepted Answer

The OpenAI MCP Server is a standard stdio Model Context Protocol (MCP) server that exposes the entire OpenAI API surface as 42 standardized tools. This allows MCP clients like Claude Code, Cursor, and Windsurf to seamlessly interact with various OpenAI functionalities.

Question 2

Which OpenAI APIs are supported by this server?

Accepted Answer

It provides comprehensive coverage across 42 tools, including OpenAI Responses, Chat Completions, Images (generation & editing), Audio (TTS, transcription, translation), Embeddings, Moderation, Files, Batch jobs, Vector Stores, Fine-tuning, and robust Large Uploads.

Question 3

What security features are built into the OpenAI MCP Server?

Accepted Answer

The server includes several security measures such as a path deny-list, write-basename deny, blocking of hosted tools (`mcp`/`shell`/`computer_use`), destructive-action gating, 401 cache invalidation, and an SSRF guard for image URLs, ensuring safer interactions with OpenAI APIs.

Question 4

How do I install and configure the OpenAI MCP Server?

Accepted Answer

You can install it globally via npm (`npm install -g openai-mcp-server`) or from source on GitHub. Configuration involves setting your OpenAI API key, typically via an environment variable (`OPENAI_API_KEY`) or a file reference, and then registering the server within your MCP client's settings (e.g., `~/.claude.json`).

Question 5

How does the server manage OpenAI API keys?

Accepted Answer

It offers flexible API key resolution, prioritizing the `OPENAI_API_KEY` environment variable, followed by a key read from a file specified by `OPENAI_API_KEY_FILE`. As a fallback, it can also retrieve API keys from Google Cloud Secret Manager if configured.

OpenAI

OpenAI

主な機能

ユースケース

主な機能

ユースケース