efirewall-ai

AI-driven L4-L7 enforcement layer for efirewall. It turns the advisory block-at-edge verdicts into an auditable blocklist and (in a later phase) drives the Traefik edge and Palo perimeter.

Autonomy mode

auditsafe

Evaluating and recording decisions only -- nothing is blocked. This is the safe default.

Confidence threshold

0.8

Block TTL (hours)

Per-run budget