Automation runs

Every time an Automation runs, on its own or as a step in a workflow, OpsLevel records the run. All Automation runs appear under Runs.

Run an Automation on its own

1. Open the Automation and start a run.
2. Choose what it runs against, such as the components (and their repositories) to change.
3. Add any plan or extra instructions for this run.
4. Start the run.

When an Automation runs as a workflow step, the workflow supplies these values, so there's nothing to fill in.

The Runs list

Runs lists recent Automation runs across all Automations, with each run's Automation, status, what it ran against, who or what started it, and when.

Run details

Select a run to see:

• What it ran against: each component, with its own status
• The result: for prompt Automations, the response; for coding Automations, the plan and the changes made in each repository
• Pull requests opened, with links to each one
• Logs, for following what the agent did or debugging a failure
• The workflow run that started it, if it ran as a workflow step

Coding Automations

A coding Automation works on each component separately, so one run can succeed for some components and fail for others. Check each component's status and logs to see where it got stuck, for example a failing build or test.

Pull requests go through your repositories' normal review process, branch protections, and required checks.

Usage

Each run shows the OpsLevel credits it used. Runs that use your own Anthropic API key don't use credits.