SignalGuardPre-action social engineering analysisHow it works

SOCIAL ENGINEERING · PRE-ACTION ANALYSIS

Understand the ask before you act.

SignalGuard breaks suspicious messages into requested actions, assets at risk, concrete security evidence and safe verification steps — without opening the links inside them.

Links are never opened Messages are not stored AI does not make the final decision
LOCAL EVIDENCE ENGINE READY
MAX 20,000 CHARACTERS

Paste a suspicious message

Email, text, direct message, payment request, or account alert.

We never open links found in your message.0 / 20,000
TRY AN EXAMPLE
Your message is not stored. Contextual analysis may be processed by the configured AI provider when enabled.

METHOD, NOT MYSTIQUE

How SignalGuard works

Evidence stays separate from contextual judgment, and neither is presented as certainty.

01

Concrete checks

Pure TypeScript rules identify requested actions, sensitive assets, manipulation patterns, and URL structure.

02

Optional context

When configured, bounded AI interprets nuance and ambiguity. The pasted message is treated only as untrusted data.

03

Deterministic policy

Application code combines evidence into a risk category. AI cannot set or override the final classification.

04

Independent verification

Action-specific guidance creates a safer route that does not depend on the communication being assessed.

Important limitation: SignalGuard cannot authenticate the sender, inspect mail headers from pasted text, or guarantee whether a communication is legitimate.