Remote MCP or REST
ChatGPT’s connector UI takes URL + Bearer — not a full JSON paste. Homepage Ask agent fills those fields into a prompt for you.
Try with your agent
Built for ChatGPT
Design checkup inside ChatGPT
Add ReviseMy as a remote MCP connector (or Custom GPT Action for REST). Use the MCP URL and Bearer try token from the homepage — then ask ChatGPT to run create_review and follow next_action.
The problem
ChatGPT can critique UI in text, but it cannot leave structured marks on pixels or wait for a human approve / request-changes gate. You need a connector that hands off to a real review link.
How ReviseMy fits
Get a try token on the homepage, add the MCP URL and Authorization: Bearer token in ChatGPT Connectors (or Actions for REST), then paste a checkup prompt. When the host supports MCP Apps the review can render inline; otherwise ChatGPT shares the review_url.
What you get with ChatGPT
Inline when available, link otherwise
If the host supports MCP Apps, the review can open in chat. Otherwise you get a review_url — same loop either way.
Same tools as every host
create_review, get_review, next_action — host-agnostic payloads so you are not locked to ChatGPT.
Quick setup
- Get a free try token on the homepage
- ChatGPT → Settings → Connectors (or Custom GPT → Actions for REST)
- Paste MCP URL + Authorization: Bearer {try_token}
- Ask it to run create_review and follow next_action
FAQ
Where is the paste-ready config?
See the ChatGPT section on /connectors#chatgpt, or use Ask agent on the homepage after you create a try token.
Does ChatGPT support MCP Apps inline?
When the host supports MCP Apps, yes. Otherwise the agent shares review_url. Details on /mcp-apps.
Full paste-ready setup for ChatGPT: Connectors · ChatGPT · MCP Apps vs review_url