What is SentinelRN?

SentinelRN is an open-source Application Trust Runtime (ATR) for React Native. It helps production mobile apps continuously evaluate runtime trust, assess AI input risk, and make explainable security decisions before sensitive actions occur.

The problem

Today's React Native security ecosystem is fragmented. You install one package for jailbreak detection, another for secure storage, another for SSL pinning, another for PII detection. Each answers a different question — and none answer the one apps actually care about:

Can I trust this environment enough to perform this action?

Most libraries also return bare booleans:

const rooted = await isRooted(); // …now what?

A boolean doesn't tell you how severe the situation is, how confident the detection is, or what your app should do about it.

The SentinelRN approach

SentinelRN returns structured, explainable reports instead of booleans:

{
  riskLevel: "high",
  score: 82,
  compromised: true,
  recommendedAction: "block_sensitive_action",
  signals: [
    {
      type: "jailbreak",
      severity: "high",
      confidence: "medium",
      platform: "ios",
      message: "Suspicious jailbreak filesystem path detected"
    }
  ]
}

Every result explains what happened, why it matters, how severe it is, how confident the detection is, and what action to consider.

Two pillars

1. Runtime integrity — identify rooted, jailbroken, emulated, debugged, hooked, or tampered device environments.
2. AI input protection — detect and redact secrets, PII, and prompt-injection attempts before user content leaves the device.

Design principles

• Risk-based. Detection is probabilistic. SentinelRN helps you make better decisions; it never claims perfect protection.
• Explainable. No bare booleans. Severity, confidence, and a recommended action on every result.
• AI is first-class. Prompt protection lives inside the runtime.
• Secure defaults. The safest path is the easiest path.
• React Native first. TypeScript, hooks, providers, Expo, and bare RN.

What it is not

SentinelRN is not an authentication framework, an MDM replacement, a backend authorization layer, or an AI provider SDK. It is a client-side signal that reduces risk — your server must still enforce trust. See the threat model.

Next: Getting Started.