Threat model & defensive controls
Every submitted URL is treated as untrusted input. The acquisition layer is deliberately defensive.
SSRF
Private, loopback, link-local and multicast destinations are blocked after DNS resolution.
Redirect abuse
Redirects are manual, limited and validated at every hop.
Resource exhaustion
15-second timeout and 900,000-byte response ceiling.
Abuse
12 requests per minute per observed client key.
Control sequence
INPUT → SCHEME → CREDENTIALS → PORT → DNS → PUBLIC IP → FETCH → SIZE → PARSE → ANALYZE
Security evidence
Every live analysis returns a security audit alongside the score, so an evaluator can inspect which defensive controls were applied.
Explicit boundaries
- HTTPS is not treated as proof of credibility.
- No coverage is not treated as proof of falsehood.
- The browser is not presented as having raw host-network visibility.