Monitoring

Agent clients ship constantly, and behaviour that passed last month quietly breaks. Monitoring re-runs your completed audit's exact plan on the cadence you choose, at reduced volume, and emails you only when something that passed before regresses. No news is the product working.

Cadences

What you control

• The cadence — change it any time; an on-demand run a check now is included.
• The alert threshold — low / medium / high severity gates the regression email, so a minor wobble doesn't page you.
• The recipient — route alerts to a different address (your on-call, a shared inbox) than the account email.
• Pause / resume / cancel — self-serve from the monitoring page, no email to support.
• The monthly digest — optional proof-of-life when everything stayed green; off if you only want to hear about problems.

What a regression report contains

The diff that matters: which scenarios regressed from the baseline, at what severity, with the same evidence standard as the audit itself — traces behind every claim. Your monitoring dashboard keeps the tick history, so “when did this start failing?” has an answer.

Monitoring pairs naturally with the Fix Loop: the regression email tells you what broke, and your coding agent pulls the details and fixes it with the audit's context already loaded.