Module threat 

Content scanning for durable memory entries.

scan matches memory content against a fixed set of patterns for content that should not flow through durable memory unreviewed: shell commands that embed credential variables, requests that carry credential material in URLs, reads of well-known credential files, instructions that rewrite agent instruction files, and inline secret or private-key material.

Each pattern is tagged with the minimum Scope that activates it: On patterns run in both modes; Strict patterns add broader, higher-recall matches that only run when the operator opts into strict mode.

Structs

Finding

A single scan match: the category, the matched text, and its location.

Enums

Scope

Active scan scope, derived from [memory.policy].threat_scan.

ThreatKind

Category of a scan match.

Functions

scan

Scan content and return every pattern match active under scope.