Review agent team runs, metrics, and quality

Review run history

The Tests and runs tab lists team coordination activity. Each item identifies its source, status, participants, and execution time.

Select a run to inspect the request, collaboration route, final response, and technical details.

Available metrics

The team summary includes total and resolved runs, average quality, average response time, accumulated credits and cost, and participating agents.

Use trends and comparable groups. A single average can hide the difference between simple requests and exhaustive coordination.

Analyze a slow or expensive run

Review how many specialists participated, whether they ran in parallel, which delegation took longest, whether all participants were necessary, and whether the profile matched the task.

Removing unnecessary participation is usually more effective than shortening every instruction indiscriminately.

Rate quality

Studio lets you rate a completed response from one to five stars. Human ratings complement automatic checks for completion and output contract compliance.

Apply consistent criteria such as accuracy, coverage, clarity, and usefulness.

Keep traceability

Selection, delegation, synthesis, failures, and cancellation are recorded. Include the run identifier when contacting support so the exact case can be located without sharing credentials.