Skip to main content

Detectors and the Detector Catalog

Detectors are the building blocks of Data Governance: each one recognizes a single type of sensitive data, such as a credit card number, API key, email address, or medical record number. You choose detectors when you create a policy.

Open Data Governance → Detectors to understand what the gateway can detect. The page has three views:

ViewPurpose
Your coverageShows detectors referenced by your enabled policies
CatalogBrowses all shipped, custom, and composite detectors
Catalog managementManages the active detector bundle version; available to administrators

Your coverage

Use Your coverage to answer: What is my organization actively protecting right now?

This view is policy-driven. A detector appears here when an enabled policy uses it. Follow See policies links to find the policies responsible for that coverage.

Catalog

Use Catalog to:

  • Browse detectors by coverage area, regulation, industry, or region
  • Open a detector to inspect how it works
  • Review custom and composite detectors
  • Test sample text against text-pattern detectors in the active catalog

The in-browser test evaluates text-pattern detectors. AI-entity detectors are evaluated by the gateway at scan time and are not included in that local test.

Custom and composite detectors

Security administrators can build detectors for organization-specific data.

A custom detector can use text patterns, word lists, context words, confidence settings, and exclusions. A composite detector represents a meaningful class that depends on multiple findings occurring together—for example, a subscriber number together with billing information.

When creating a composite detector, you can define:

  • Findings that must all be present
  • Groups where at least one finding must be present
  • Findings that must not be present
  • A minimum number of distinct findings
  • A minimum confidence
  • The synthesized name reported when the combination matches
tip

Test custom logic with representative positive and negative samples before adding it to a blocking policy.