OpenAI product questions age quickly when Codex can only use model context or old repo notes. Adding the OpenAI Docs MCP server gives a new Codex session a named Model Context Protocol connection to OpenAI developer documentation, so API, Apps SDK, and Codex answers can be pulled from the public docs server while work is underway.
OpenAI hosts the Docs MCP server at https://developers.openai.com/mcp as a public streamable HTTP endpoint. The official quickstart uses openaiDeveloperDocs as the server name, and Codex shares MCP configuration between the CLI and IDE extension through config.toml.
The endpoint is documentation-only and does not call the OpenAI API for the session. Save only the endpoint URL, restart or reopen any already-running Codex session, and ask for the OpenAI developer documentation MCP server explicitly unless the project already has durable AGENTS.md guidance for OpenAI lookups.
Related: How to add a URL-based MCP server to Codex
Related: How to list Codex MCP servers
Related: How to get Codex MCP server details
Related: How to remove a Codex MCP server
$ codex mcp add openaiDeveloperDocs --url https://developers.openai.com/mcp Added global MCP server 'openaiDeveloperDocs'.
Save the endpoint exactly as published. Do not append API keys, bearer tokens, cookies, or query parameters to the URL.
$ codex mcp get openaiDeveloperDocs --json
{
"name": "openaiDeveloperDocs",
"enabled": true,
"disabled_reason": null,
"transport": {
"type": "streamable_http",
"url": "https://developers.openai.com/mcp",
"bearer_token_env_var": null,
"http_headers": null,
"env_http_headers": null
},
"enabled_tools": null,
"disabled_tools": null,
"startup_timeout_sec": null,
"tool_timeout_sec": null
}
Confirm transport.type is streamable_http and transport.url is https://developers.openai.com/mcp.
Related: How to get Codex MCP server details
$ codex mcp list Name Url Bearer Token Env Var Status Auth openaiDeveloperDocs https://developers.openai.com/mcp - enabled Unsupported
Unsupported under Auth is expected for this public endpoint because there is no bearer-token or OAuth login flow.
Related: How to list Codex MCP servers
$ codex
Run /mcp inside the Codex terminal UI and check that openaiDeveloperDocs appears. Restart already-open CLI or IDE sessions so they reload the saved MCP configuration.
Use the OpenAI developer documentation MCP server to find the current Responses API tools request schema.
Allow the read-only MCP tool call if Codex asks for approval. Add a narrow AGENTS.md rule only when the project regularly works on OpenAI APIs, Apps SDK, or Codex behavior.