PhishLens

Threat Model

Protected Assets

Risks

Assumptions

Possible Abuse

Mitigations

Accepted Risks And Limitations

SSRF via DNS name resolution. Blocking IP literals at request time without resolving DNS names is a deliberate trade-off. Resolving every submitted hostname before connecting would add latency, require a DNS dependency on the backend, and introduce a TOCTOU window between the resolution check and the actual connection. The accepted risk is that an attacker-controlled hostname resolving to a private address could reach internal services via the TLS inspection path. This is mitigated by the fact that the backend only makes a TLS handshake (no HTTP request body is sent to the target), the connection is limited to port 443, and the backend runs as a non-root user in a container. A network-level egress firewall blocking RFC 1918 ranges is the recommended control for production deployments.