Automation

AzureScout is built to run without a human at the keyboard: no interactive prompts, no COM dependencies, and a run log written for every run including failed ones.

Pick a host

Page	Use it when
Azure Automation Account	You want a scheduled runbook inside Azure, authenticating with a managed identity and writing to blob storage
GitHub Actions	Your pipelines already live in GitHub — a composite action is provided
Azure DevOps	Your pipelines live in ADO, or you also want to inventory ADO itself: projects, pipelines, service connections, repos and agent pools

What every unattended run needs

• A non-interactive identity — a managed identity, or a service principal with a secret or certificate. See Authentication.
• Reader at the scope you want covered. Anything narrower silently reduces what is collected, so run -PermissionAudit once from the same identity first.
• Somewhere durable for the output. In an Azure Automation sandbox the filesystem is discarded when the job ends, so blob storage is the only copy that survives.

Reading a headless run

Every run writes scout-run.log with phases, elapsed times, counts and full error detail — including runs that failed. Two further artefacts make a run auditable rather than merely finished:

• raw-inventory.json — everything collected, written before any manifest decided what to display, so a resource type no collector claims still leaves a trace.
• collector-rowcounts.json — what each collector produced, and why it produced nothing.

That last distinction is the one that matters unattended: an empty worksheet and a collector that was never permitted to run look identical in a report, and these files tell them apart.