Skip to main content

How it works

Endpoint monitoring re-tests a project's known findings against its live target on a schedule you set. Each re-test sets the finding to fixed when the issue no longer reproduces, or to not fixed when it still does.

Which findings are monitored​

AISafe enrolls findings automatically. A finding is monitored when it comes from a black-box or white-box assessment on the project, has been confirmed, carries a proof-of-concept or reproduction steps, and is open, not fixed or fixed. You do not need to mark anything first: an open finding is re-tested to see whether it has been fixed, and a fixed finding is re-tested to catch a regression.

A project monitors at most 50 findings. When more qualify, the ones left out have the lowest severity and were discovered most recently. The list of monitored findings is refreshed each time you save the project's monitoring settings, and the settings panel shows how many more findings will be added at the next save.

What a monitoring run does​

  1. When the assessment that found a finding stored a live test plan, the run replays that plan against your target. A replay sends the same requests the assessment sent and needs no AI model.
  2. Each monitored finding that no plan settled is re-tested by an agent, which follows the finding's proof-of-concept or reproduction steps against your target.
  3. Each finding gets its result. If the issue still reproduces, the finding becomes not fixed. If it no longer reproduces, the finding becomes fixed. The finding keeps its existing finding code either way. A finding you marked false positive, accepted risk or duplicate keeps that status, and the latest re-test result is recorded beside it.

When a check is blocked by the network or cannot run, AISafe records the result as inconclusive and leaves the finding's status unchanged, so an unreachable target never appears secure.

What counts as a regression​

A regression is a fixed finding that reproduces again. This can happen when:

  • A later code change reverts or bypasses the fix.
  • A dependency update brings back a vulnerability that an earlier release patched.
  • An infrastructure change exposes an attack that was mitigated before.

When monitoring finds a regression, AISafe:

  1. Changes the finding to not fixed.
  2. Raises a monitoring.regression event, which reaches your outbound webhook subscriptions.
  3. Notifies you, including in Slack if you have connected it.

Open the finding to see the regression details, find the cause and fix it. See Monitor for regressions.

Cost​

Each monitoring run is one unit of work, whatever number of findings it re-tests. A Pro plan includes a monthly number of monitor runs; after that, each run costs $2 in credits. See Credits and billing.