Agent runs and transcripts - Amnify

Every piece of work Amnify does happens in a run, and every run keeps its complete transcript. An agent that reaches a conclusion has to show the steps that got it there.

Trigger	What it is
Full security scan	A complete run across every selected scope. The first one is your initial scan
Scan automation	A scan of one scope, from a schedule, a push, a new advisory, or a manual start
Penetration test	An authorized test against a target
Pull request review	A review of one pull request
Context answer	Re-evaluation of findings after you answered a question
Remediation	Planning or executing a fix
Conversation	Your chat with Amy

The transcript

Open a run to read what happened step by step: what was read, which checks ran, what was found, what was decided, and what was written. Specialist work is nested under the run that dispatched it. An incomplete run can be stopped from its page. Everything it wrote before stopping is already saved.

What runs when

Amnify never runs two conflicting jobs against the same scope, so a scan and a penetration test never collide. Different scopes run at the same time, and a pull request review is never held up by other work.