Expected outcome
ChatGPT or an OpenAI client can perform the agreed job in the selected workspace without receiving broader access than it needs.
Tempera is in private preview and provides private APIs across its product areas. Access is arranged through onboarding for an agreed evaluation, with product-specific credentials, endpoints, and environment details supplied directly to your team. API maturity ranges from preview to research, and no shared cross-product endpoint is implied.
Before you start
- A provisioned design-partner workspace
- An approved ChatGPT or an OpenAI client use case
- A workspace owner who can review requested permissions
Steps
- 01
Confirm the job and workspace
Write down what ChatGPT or an OpenAI client needs to do and which staging workspace contains the permitted data.
- 02
Use the supplied connection details
Use the client registration, sign-in link, or configuration supplied during onboarding. Tempera does not publish a shared endpoint or generic public configuration in this guide.
- 03
Match permissions to this job
Approve the named tools and trace reading needed for this review. Keep trace writing, dataset access, evaluation execution, and administrator access out of the connection.
- 04
Run the client-specific check
Open one known staging trace through the approved tool and summarize the evidence for its decision point without modifying it.
What success looks like
- ChatGPT or an OpenAI client shows as connected for the selected workspace.
- The approved permissions match the documented job.
- The OpenAI client can inspect the intended trace through approved tools without write access.
If something is blocked
- If the connection is expired or revoked, reconnect the same client instead of creating a broader credential.
- If a permission is missing, request only that permission and repeat the narrow check.
- If the wrong workspace appears, stop before running a tool and reconnect from the intended workspace.