Software factory / Runs
Factory runs
An agent ran on a pull request, an issue or the default branch, and you want to know what it did. Or a run failed, waits or goes wrong, and you want to act on it.
2 min read
Every run of an agent has a log. A run that failed says why, and what to do next. Pick your task:
| You want to | Read |
|---|---|
| Find a run, and read what its agent did | Find and read a run |
| Know which agent workflow opened a pull request, and what its agent did | Find the run behind an agent's pull request |
| Know why a run failed, and fix the cause | Why a run failed |
| Start a failed run again | Retry a run |
| Know why an agent waits or will not run | Why an agent waits |
| Stop a run that is wrong | Stop a run |
To start an agent that did not run yet, see Start an agent run.
Who can do what
- People with write access read the list of runs and the logs. They read the exact reason of a failure, retry runs and stop runs.
- The author of the issue or the pull request reads the exact reason of a failure. The author can retry a run that failed for a temporary cause.
- Everyone who can read the item sees which runs are queued or running, and a general reason for a failure.
Where runs show
- The Factory panel of each issue and pull request. It updates by itself when a run is added, starts or ends.
- The runs page of the repository: Agents, then Runs (
/<owner>/<repo>/agents/runs). - The issues list, the pull requests list and the header of the item: a blue badge while a run is queued or running.