audit echo vs log across all scripts — outcomes always visible, verbose for per-item loops

claude-data/projects/-root/memory/project_watchdog_health_urls.md

@@ -0,0 +1,19 @@
+---
+name: project-watchdog-health-urls
+description: Planned improvement — add HTTP health check URLs for auth stack containers in HOST1_WATCHDOG_CONTAINER_URLS
+metadata: 
+  node_type: memory
+  type: project
+  originSessionId: ffec43cd-13e3-4911-878f-40459f7d16a9
+---
+
+Add health check URLs for critical containers that are currently only checked via `docker ps` (alive but hung won't be caught).
+
+**Why:** A hung-but-not-crashed NPM/Authelia fails all traffic silently with no watchdog trigger. Currently only Emby has a health URL.
+
+**How to apply:** When user is ready, add to host1.conf:
+- Define `HOST1_NPM_ADMIN_URL="http://localhost:7818"` near the NPM identity section (no existing URL var for admin panel — `HOST1_NETWORK_WATCHDOG_NPM_URL` is the external domain, different thing)
+- Add to `HOST1_WATCHDOG_CONTAINER_URLS`: NPM, Authelia (9091), Authelia-Secondary (9092), Lldap-Gmer4Lfe (17170) — confirm ports before adding
+- Pattern: define URL var, reference it in the array (same as HOST1_EMBY_URL pattern)
+
+NPM admin port confirmed by user: **7818** (not 81 — that's the partnership WebUI port in HOST1_PARTNERSHIP_AUTH_WEBUIS)