Guide
Campaign cadence
When to run a full scan, an incremental, and a retest
The commands live on Campaigns. This page is the rhythm.
A simple loop
- Full scan when you first enroll a repo, after a methodology change, and on a release or monthly boundary. Triage it before you call it a baseline.
- Incremental on pull requests (or any branch update). The webhook or API supplies changed paths; a daemon on a box you control runs the models.
- Retest each serious fix before you ship, from the recorded fix commit.
Keep one finished campaign as the comparison baseline while a release train is in flight. Do not compare incomplete campaigns as if missing findings meant "fixed." Pending slices make the gate pending.
Reading the comparison
The selected campaign is current. Its baseline is the one you named, or the previous campaign.
- New: seen now, not in the baseline.
- Resolved: seen in the baseline, not now.
- Carried forward: open in both.
Wording drift that de-duplicates to the same finding counts as carried forward, not new.