API Reference

The primary entry point is the cohesive SentinelRN object.

import { SentinelRN } from "@sentinelrn/core";

SentinelRN.configure(config)

Apply configuration. Optional — secure defaults apply otherwise.

SentinelRN.configure({
  policy: "monitor",
  integrity: { includeEvidence: false },
  ai: { includeFindings: false, redactMatches: true },
});

Option	Type	Default	Description
policy	SentinelPolicy	"monitor"	Default policy across modules.
integrity.includeEvidence	boolean	false	Attach raw evidence to signals (may be sensitive).
ai.includeFindings	boolean	false	Include the matched substring on findings.
ai.redactMatches	boolean	true	Populate redacted previews on findings.

SentinelRN.integrity

Member	Signature	Description
check()	() => Promise<ThreatReport>	Run an integrity check via the registered provider.
registerProvider(p)	(IntegrityProvider) => void	Register the active platform provider.
getProvider()	() => IntegrityProvider | undefined	The current provider.

SentinelRN.ai

Member	Signature	Description
guardPrompt(args)	(GuardPromptArgs) => AIGuardResult	Inspect AI-bound input under a policy.

type GuardPromptArgs = { input: string; policy?: "monitor" | "warn" | "block" | "strict" };

SentinelRN.policy

Member	Signature
evaluateIntegrity(report, policy?)	(ThreatReport, SentinelPolicy?) => PolicyDecision
evaluateAI(result, policy?)	(AIGuardResult, SentinelPolicy?) => PolicyDecision

SentinelRN.redaction

Member	Signature	Description
redact(input)	(string) => string	Replace secrets/PII with placeholders.
inspectAndRedact(input)	(string) => RedactionResult	Sanitized text plus findings.

SentinelRN.registerIntegrityProvider(provider)

Shortcut for integrity.registerProvider. Used by @sentinelrn/native.

SentinelRN.version

The package version string.

Types

type RiskLevel = "low" | "medium" | "high" | "critical";
type Severity = "low" | "medium" | "high" | "critical";
type Confidence = "low" | "medium" | "high";
type RecommendedAction =
  | "allow" | "monitor" | "warn_user" | "block_sensitive_action" | "block_session";

type SentinelPolicy = "monitor" | "warn" | "block" | PolicyConfig;

interface PolicyConfig {
  mode?: "monitor" | "warn" | "block";
  blockOnHighRiskDevice?: boolean;
  redactSensitiveInput?: boolean;
  blockPromptInjection?: boolean;
  allowInSimulator?: boolean;
}

interface IntegrityProvider {
  readonly name: string;
  collectSignals(): Promise<RawSignal[]> | RawSignal[];
}

Advanced exports

For extension and advanced use, @sentinelrn/core also exports the underlying engines: assessRisk, aggregateScore, riskLevelFromScore, buildThreatReport, normalizeSignal, detectFindings, guardPrompt, redact, inspectAndRedact, evaluateIntegrity, evaluateAI, resolvePolicy, createSentinel, and the detector/rule tables (PATTERN_DETECTORS, INJECTION_RULES).

Errors

The SDK never crashes the host app because a detector fails. Errors are structured:

class SentinelError {
  code: string;
  module: "integrity" | "ai" | "policy" | "redaction" | "native" | "core";
  message: string;
  cause?: unknown;
}