Suspicious-link inspection
Flags risky URL structures such as IP-literal hosts, embedded user-info, punycode/IDN indicators, invalid ports and encoded path concerns. This is not live URL detonation or a guarantee that a link is safe.
Statera family · Intelligent email security & AI assistant
AegiDex Sentinel is being built to help identify suspicious email signals, explain why a message deserves attention, and support safer review and organization without pretending to provide device-wide or operating-system security.
Current release state: pre-production. Live mailbox access remains disabled until trusted identity, consent, provider, and security gates are satisfied.
What the MVP already implements
Flags risky URL structures such as IP-literal hosts, embedded user-info, punycode/IDN indicators, invalid ports and encoded path concerns. This is not live URL detonation or a guarantee that a link is safe.
Produces deterministic findings with evidence and limitations instead of silently declaring a message clean when no heuristic fires.
Evaluates file size/type metadata without claiming malware scanning before the isolated antivirus runtime exists.
Tenant/account/provider scope, consent structure and replay protections are modeled fail-closed. Production issuer verification remains a release gate.
User-reported threats can be organized into explainable triage outcomes that preserve a human-review state rather than fabricating final authority.
Aggregate risk summaries are designed to reduce unnecessary message-content retention while preserving useful security context.
AI assistant
AegiDex can support explanation, organization suggestions and support escalation, but AI output is not allowed to grant permission, override security controls, or execute destructive mailbox changes by itself.
Release readiness