App-level persistent IP ban for Clusev's own login (distinct from the remote-fleet
Fail2banService): BannedIp model, BruteforceGuard service, guests-only BlockBannedIp
middleware, Login/2FA hooks + audit, a new Anmeldeschutz settings tab, clusev:unban CLI.
Hardened after an adversarial multi-lens review: documented IP trust model (Caddy sole
proxy, prod-only trustProxies), guests-only enforcement so an authed operator is never
locked out and can self-unban, current-IP self-whitelist, concrete cache strategy,
inet_pton CIDR matching, R5 confirm modals, errors.php for the 403 strings.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>