ThreatReport

Report describing a detected threat span.

NOTE: A ThreatReport represents a span, not a single character. Adjacent suspicious characters should be grouped into one report.

Properties

decodedPayload?

optional decodedPayload: string

Defined in: types.ts:168

Optional decoded payload extracted from concealed content.

Example: "ignore previous instructions"

loc

loc: ThreatLoc

Defined in: types.ts:137

Location of the threat start

message

message: string

Defined in: types.ts:134

Human-readable diagnostic message.

Describes WHAT was detected and WHY it matters. This should not include remediation steps.

offendingText

offendingText: string

Defined in: types.ts:145

The substring responsible for the detection.

This may contain multiple characters if the threat represents a sequence.

readableLabel?

optional readableLabel: string

Defined in: types.ts:153

Optional readable label for UI rendering.

Example: "[ZWSP × 3]"

referenceUrl

referenceUrl: string

Defined in: types.ts:176

Reference documentation explaining the risk.

Example: https://promptshield.js.org/docs/detectors/invisible-chars#PSU001

ruleId

ruleId: string

Defined in: types.ts:120

Stable rule identifier.

Example: "PSU001", "PST001", "PSI002"

severity

severity: Severity

Defined in: types.ts:126

Risk severity

suggestion?

optional suggestion: string

Defined in: types.ts:160

Suggested remediation guidance.

This is optional and may vary by environment (editor, CI, UI).

suppressed?

optional suppressed: boolean

Defined in: types.ts:182

Indicates whether this threat was suppressed by an ignore directive.