Skip to Content
FlowOverview

Flow

This is the journey of a scan: what happens between pressing “scan” and seeing findings, the statuses you’ll see along the way, and how Classifyre keeps results accurate as the same source is scanned again and again.

It’s the connective tissue between Sources, Detectors, and Investigations.


A source and its scans

A source is a connection you’ve configured; a scan is one run against it. A source can be scanned many times — manually or on a schedule — and each run is recorded so you can see its progress and history.

At any moment a source shows a simple status reflecting its most recent run:

StatusWhat it means
PendingThe source exists but isn’t scanning right now (never scanned, or its last run finished).
RunningA scan is in progress. Only one scan runs per source at a time.
CompletedThe last scan finished and every item processed cleanly.
WarningThe last scan finished, but some items couldn’t be processed (check the run logs).
ErrorThe last scan failed partway through.

What happens during a scan

When a scan runs, it moves through four phases you can watch in the app:

PhaseWhat’s happening
TriggeredThe scan is queued and starts as soon as there’s capacity.
DiscoveringThe source is listed — every page, file, or record — so progress can be shown (“134 of 500 scanned”).
ScanningEach item is read as an asset, its content extracted, and detectors run over it. Findings start appearing.
CompletingThe run wraps up: items that have disappeared from the source are reconciled, and findings are brought up to date (see below).

How much of a source a scan reads, and in what order, is governed by its sampling strategy.


What repeat scans do

Classifyre remembers what it saw last time, so a second scan doesn’t start from scratch or pile up duplicates. Every item is compared to the previous run and sorted into one of four buckets:

OutcomeWhat Classifyre does
NewReads it and records any findings as freshly detected.
UpdatedRe-reads it and refreshes its findings.
UnchangedNotes it’s still there, without redoing the work.
DeletedMarks it gone and auto-resolves the findings that came from it.

This is what makes results trustworthy over time: the same issue is tracked, not re-reported, and issues that genuinely go away are closed automatically.


Findings stay up to date

Because items are reconciled each run, findings persist and update rather than duplicate. A finding that’s still there is kept; one that’s been fixed at the source is automatically resolved; one that comes back is reopened.

Your own decisions are always respected. If you’ve marked a finding as a false positive, ignored, or resolved, a later scan won’t silently overturn that.

The full anatomy of a finding — its fields, severity, confidence, and lifecycle — lives in Findings & Results.


What happens to findings next

Producing findings is where this lifecycle ends and the investigation layer begins. From here, findings are watched, connected, and worked:

  • Investigations — the full picture of how findings become inquiries, fingerprints, cases, and hypotheses.
  • Inquiries — standing questions that keep surfacing matching findings.
  • Fingerprints — duplicate and similarity detection across assets.
  • Autopilot — AI agents that act on new findings automatically after every scan.
Last updated on