Core concepts
The vocabulary you need — scans, status, tags, severity, classification, evidence, campaigns, runs, policies, and alerts.
A quick tour of the terms used throughout Kaminari Ad and this documentation.
Scan
A scan is a single verification of one URL (or ad tag) in one context (country + device emulator + proxy). It is the atomic unit of work. Every scan moves through a lifecycle and produces tags, a classification, and evidence.
Status
A scan's status tracks where it is in its lifecycle: pending ->
running -> crawled -> checking -> completed (or partial when
some landings failed, failed, or cancelled). Re-running a finished
scan moves it through rechecking, and a scan granted another crawl
attempt returns from running to pending, so the chain is not
strictly forward-only. Status describes progress, not a safety score —
the findings themselves are the tags.
Tags
A tag is a labelled finding attached to a scan — for example an
auto-redirect, a malware hit, or a custom rule match. Each tag has a
category (system source, AI, behavioural, creative, custom…) and a
severity — high, medium, or low. Tags come from system
detectors, your custom rules, and the AI classifier, and they are the
signal you alert and automate on.
Classification
Beyond tags, every scan is classified: the detected brand, an AI content category, the canonical IAB v3 category, and any custom taxonomy assignments. Classification answers "what kind of ad is this?"; tags answer "what is wrong with it?".
Evidence
Evidence is everything the crawler captured: the rendered screenshot of the creative and each landing page, the full redirect chain (every hop from the creative to the final URL), and the raw page artifacts. Every tag is grounded in this evidence, so a finding is always defensible.
Campaign and run
A campaign is a reusable definition of what to scan and how (URLs or ad tags, countries, emulators, schedule, policy set). Each time a campaign executes it creates a run — a batch of scans. Campaigns can be grouped for bulk operations.
Policy set
A policy set encodes the rules a campaign's results are judged against — which tags or categories matter and what raises an alert. Policy sets can be private to your org or submitted for moderation.
Alert
An alert is raised when a scan's tags cross a threshold defined by your policy set. Alerts have a lifecycle — open, escalated, resolved, or dismissed — and can be delivered to Slack, Telegram, or email.