Inspect Run Details
Use the run page to review event logs, responses, artifacts, terminal access, and sandbox lifecycle controls.
The run detail page is the best place to answer the question, "What actually happened during this run?"
The Main Areas Of The Run Page
The run page includes four major investigation surfaces:
- Event log
- Artifacts
- Responses
- Terminal
It also shows run metadata such as duration, snapshot, sandbox size, usage, and current sandbox state.
Event Log
The event log is the first place to look when a run seems wrong.
It helps you answer:
- when the response started
- whether tools were called
- where a failure occurred
- whether the runtime reached a terminal state
Use the severity filters when you only want errors or warnings. If you need more detail, expand an event to inspect the structured payload.
Artifacts
Artifacts are files created during the run. Use this tab when the output is a generated document, report, transformed file, or another deliverable that should be downloaded after completion.
If a run should have produced a file but none appears, check the event log and response tabs next.
Responses
The responses tab shows the response objects associated with the run, including status and whether they were streamed or created in the background.
This is useful when you need to compare:
- what the API returned
- whether the response completed
- whether the runtime marked the response as failed, incomplete, or cancelled
Terminal
While the sandbox is active, the terminal tab gives you live runtime access. This is often the fastest way to confirm whether the process is still working, waiting, or blocked.
When the sandbox is no longer active, the page can show that the terminal is unavailable. If permissions allow it, you may also see options to resume or stop the sandbox.
Resume And Stop Controls
If the sandbox is stopped or archived and your permissions allow sandbox management, you can resume it from the run page. If the sandbox is running, you can stop it.
Use these actions carefully:
- resuming restarts billable uptime
- stopping ends the current sandbox segment
Practical Triage Order
When a run fails or behaves strangely, this order usually works well:
- check the run status and sandbox state
- review the event log
- inspect responses
- inspect artifacts
- open the terminal if the sandbox is still active