How to add a server in ChatGPT

By mcp-guide · 0 stars

Prerequisites:

- Plus, Pro, Business, Enterprise or Edu on the web. Not Free.
- On workspace plans, an admin has enabled developer mode. Enterprise and Edu admins can also set which roles use which apps and actions.
- A remote server: public HTTPS URL (usually ending in `/mcp`) or a Secure MCP Tunnel ID. ChatGPT can't launch local processes.
- Streamable HTTP or SSE transport.

1. Open Settings > Security and login.
2. Turn on Developer mode. Older guides say Settings > Apps & Connectors > Advanced.
3. Open ChatGPT Plugins and click "+".
4. Enter a name and a description. The model reads the description when deciding whether to use the server.
5. Enter the endpoint URL or tunnel ID.
6. Choose auth: OAuth, No Authentication, or Mixed.
7. Complete the OAuth sign-in if you chose it.

When the model calls a write tool, you should see a confirmation prompt with the full JSON payload available. You can remember the choice for that conversation only.

After you change the server's tools:

1. Open ChatGPT Plugins > your connection.
2. Click Refresh.
3. Start a new chat.

Naming: in July 2026 the Plugin Directory replaced the App Directory, and MCP apps now ship inside plugins. Guides from late 2025 to mid-2026 say "apps" or "connectors".

### Sources

- https://developers.openai.com/api/docs/guides/developer-mode
- https://developers.openai.com/plugins/deploy/connect-chatgpt
- https://help.openai.com/en/articles/12584461-developer-mode-and-mcp-apps-in-chatgpt