anubis: bot-challenge proxy in front of the guarded vhosts (two-tier chain)

Bots and AI scrapers hit the SSO-fronted vhosts and eat Authentik auth
work for 403s. Anubis verifies clients first: DENY for known AI scraper
UAs, JS challenge for everyone else; verified traffic resumes the normal
Authentik forward_auth chain on Caddy's internal listener (:8016).

Chain: client -> Caddy (TLS) -> Anubis :8923 (verify) -> Caddy :8016
(auth + routing resumes; Host survives both hops) -> app. Reverse-proxy
through Anubis, not forward_auth: at the pinned tag a CHALLENGE answers
HTTP 200 + HTML, and forward_auth treats 200 as "authenticated" — the
page would pass through unchallenged. Exempt by construction (never
routed through Anubis): plex (plain HTTP legacy), auth.* (IdP login
must be reachable ungated), gitea (git smart-HTTP clients cannot solve
JS challenges), and every app with its own login.

Pinned ghcr.io/TecharoHQ/anubis:v1.13.0 — the tag was source-verified
before pinning: at this tag the policy parser is JSON-only with a
minimal bots[] schema (no imports, no CEL expressions), flags are the
config surface via flagenv (TARGET/POLICY_FNAME map to --target/
--policy-fname), and the image's exec-form HEALTHCHECK is inherited by
the quadlet unit (dodges the v1 generator's HealthCmd space-split quirk).

Rollback nuance: stopping anubis 502s everything behind the chain (the
public listeners proxy through it); a true rollback is reverting the
Caddyfile and redeploying, not just systemctl --user stop anubis.

dankclaw/anubis/policy.json

@@ -0,0 +1,15 @@
+{
+  "bots": [
+    {
+      "name": "deny AI scrapers",
+      "user_agent_regex": "(?i)^(GPTBot|ChatGPT-User|OAI-SearchBot|CCBot|anthropic-ai|ClaudeBot|Claude-Web|PerplexityBot|PerplexityUser|Google-Extended|Applebot-Extended|Bytespider|Amazonbot|meta-externalagent|DuckAssistBot|cohere-ai|YouBot|KimiBot)",
+      "action": "DENY"
+    },
+    {
+      "name": "challenge everything else",
+      "path_regex": ".",
+      "action": "CHALLENGE"
+    }
+  ],
+  "dnsbl": false
+}

dankclaw/anubis/quadlet/anubis.container

@@ -0,0 +1,50 @@
+# Quadlet unit: Anubis bot-challenge proxy (rootless, boot-persistent).
+#
+# Install path on dankclaw: ~/.config/containers/systemd/anubis.container
+# Unit name: anubis.service (container name: anubis)
+#
+# Two-tier chain: the public Caddy listeners for the guarded vhosts proxy
+# everything to this container (port 8923, internal only — no published port,
+# so metrics on :9090 stay reachable only inside dankclaw_default); Anubis
+# verifies the client (challenge/DENY per policy) and forwards accepted
+# traffic back to Caddy's internal listener (TARGET below), where the normal
+# per-vhost routing — including Authentik forward_auth — resumes. Host headers
+# survive both hops, so vhost matching on the internal listener still works.
+#
+# Version pin: ghcr.io/TecharoHQ/anubis:v1.13.0 was source-verified before
+# pinning (policy parser is JSON-only at this tag, flags are the config
+# surface via flagenv: an env var named TARGET sets --target, POLICY_FNAME
+# sets --policy-fname). Upgrading = stop unit, change Image=, deploy; policy
+# schema must be re-verified against the new tag first.
+#
+# No secrets here: Anubis holds no credentials (memory store; challenge
+# cookies are client-side JWTs). A restart invalidates issued challenges —
+# users just re-challenge once. See serverAppConfigs-docs/Infrastructure/Anubis.md.
+
+[Unit]
+Description=Anubis bot-challenge proxy container
+After=network-online.target
+Wants=network-online.target
+
+[Container]
+ContainerName=anubis
+Image=ghcr.io/TecharoHQ/anubis:v1.13.0
+Pull=never
+Network=dankclaw_default
+# flagenv maps env names to flag names (TARGET -> --target, POLICY_FNAME ->
+# --policy-fname). TARGET points back at Caddy's internal listener; the
+# container DNS name resolves on dankclaw_default.
+Environment=TARGET=http://caddy:8016
+Environment=POLICY_FNAME=/policy/policy.json
+Volume=/home/dank/serverAppConfigs/dankclaw/anubis/policy.json:/policy/policy.json:ro
+# No HealthCmd: the image ships an exec-form HEALTHCHECK
+# (["/app/bin/anubis","--healthcheck"]) and quadlet inherits it when no
+# HealthCmd is set — this dodges the v1 generator's space-splitting quirk
+# that mangled quoted probes for the authentik units.
+
+[Service]
+Restart=always
+RestartSec=5
+
+[Install]
+WantedBy=default.target