Chillframe's connector is standard MCP, so ChatGPT and Grok can run your account the same way Claude does. Both use the same server URL and the same Chillframe sign-in:
https://mcp.chillframe.com/mcp
In ChatGPT, open Settings → Connectors and, under Advanced, turn on Developer mode. OpenAI gates this by plan, and on work accounts an admin controls it. If you don't see the toggle, your plan or workspace doesn't include custom connectors yet.
Back in Connectors, create a new connector. Name it Chillframe and paste https://mcp.chillframe.com/mcp as the server URL, with OAuth as the authentication.
Sign in with your Chillframe account when the sign-in screen opens, choose the accounts the agent can act on, and click Allow. It is the same consent screen Claude uses.
ChatGPT enables custom connectors per conversation: pick Chillframe from the developer-mode tools in the composer before you ask it to do something.
Go to grok.com/connectors and choose New Connector, then Custom. xAI is rolling custom connectors out by plan; if Custom isn't offered, your plan doesn't include it yet.
Paste https://mcp.chillframe.com/mcp as the server URL and complete the sign-in: your Chillframe account, the accounts the agent can act on, Allow.
Grok discovers the tools and offers them in conversations from then on.
Whichever agent you connect, the grant is the same: it appears under Settings → Authorized agents in Chillframe with its own activity feed and Revoke button, spending and rate limits apply identically, and the hard boundaries in "What an agent can never do" hold for every client.
ChatGPT's and Grok's own menus change more often than ours. If a step here doesn't match what you see, email [email protected] and a human will read it.
