PromptShieldAPI DocsFunctions
scanInjectionPatterns()
scanInjectionPatterns(text, options?, context?): ThreatReport[]
Defined in: injection-patterns.ts:100
Scan for deterministic prompt-injection patterns.
Detection strategy:
- Scan line-by-line for stable location reporting
- Attempt direct regex detection first
- Fall back to normalized detection
Span semantics: offendingText = matched instruction phrase or entire line
Parameters
text
string
options?
ScanOptions = {}
context?
ScanContext = {}