Setup guide
How to add an MCP server to ChatGPT
ChatGPT connects to remote MCP servers as "connectors". This is for hosted (HTTP) servers — local stdio commands are not supported here.
Steps
- 1
Open Connectors
In ChatGPT go to Settings → Connectors (available on paid plans). For custom servers, enable Developer mode / custom connectors.
- 2
Add a custom connector
Choose "Add custom connector" and paste the remote MCP server URL (an https endpoint). Local command-based servers won’t work — the server must be hosted.
- 3
Authorize
Complete any OAuth / token step the server requires, then enable it for your chats.
- 4
Use it
In a chat, the connector’s tools are available to the model when relevant to your request.
FAQ
Can ChatGPT use MCP servers?
Yes — ChatGPT supports remote MCP servers as "connectors" (Settings → Connectors, paid plans, with developer/custom-connector mode for your own). It connects to hosted HTTP servers, not local stdio commands.
Why can’t I add a local MCP server to ChatGPT?
ChatGPT runs in the cloud, so it can only reach servers exposed over the internet (a remote HTTP endpoint). Local command-based servers are for desktop clients like Claude Desktop, Cursor, or VS Code.
Do I need a paid ChatGPT plan for MCP connectors?
Custom connectors are gated to paid tiers and may require enabling developer mode; availability changes, so check current Settings → Connectors.
Now pick a server to add
Browse verified MCP servers with copy-paste config for ChatGPT and other clients.
Browse connectorsOther clients