I’m using the existing Vercel MCP connector in ChatGPT. A read request for an existing project/alias returns 403 Forbidden. The connector was previously reconnected, but the read was still denied afterward.
General reproduction from the existing attempt:
- Use the connected Vercel MCP integration in ChatGPT.
- Request a read of an existing project/alias.
- Receive 403 Forbidden.
Separately, signing in to the Vercel dashboard in a browser succeeds. I understand that browser sign-in does not establish that the connector has the required authorization.
What is the supported way to diagnose whether this involves connector authorization or team permissions? Please advise what sanitized diagnostic information would be useful, without changing security settings, creating new tokens, or bypassing access restrictions. I haven’t established the cause.