Ship Herman chat, Hermes KPI explorer, and intelligence analytics.

Adds floating Herman assistant with voice, Hermes technical KPI dashboard with file explorer, and a wins/kansen analytics pulse with Hermes findings hooks. Includes optional local HTTPS via Caddy for mic permissions (public certs only).

cockpit/app/services/herman.py

@@ -379,16 +379,23 @@ async def chat(
             }
 
     backend = (settings.HERMAN_LLM_BACKEND or "auto").strip().lower()
-    if backend in ("router", "openrouter", "llm"):
+    # Direct LLM backends (Hermes Agent API, OpenRouter, …) — skip offline orchestrator.
+    if backend in ("hermes", "router", "openrouter", "llm"):
         try:
             return await _chat_via_llm_router(message, channel)
         except Exception as exc:
+            hint = (
+                "Controleer HERMES_API_KEY / Hermes Agent (:8642)."
+                if backend == "hermes"
+                else "Vul OpenRouter key in via Instellingen → AI / LLM."
+            )
             return {
                 "agent": "herman",
                 "agent_label": "Herman",
-                "reply": f"LLM provider niet beschikbaar: {exc}. Vul OpenRouter key in via Instellingen → AI / LLM.",
+                "reply": f"LLM provider niet beschikbaar: {exc}. {hint}",
             }
 
+    # orchestrator | auto — try Herman orchestrator first
     try:
         async with httpx.AsyncClient(timeout=620.0) as client:
             r = await client.post(
@@ -427,8 +434,8 @@ async def chat(
                     "agent_label": "Herman",
                     "reply": (
                         f"Orchestrator offline ({exc}). "
-                        f"Cloud LLM ook niet beschikbaar ({llm_exc}). "
-                        "Voeg OpenRouter API key toe: Instellingen → AI / LLM."
+                        f"Cloud/Hermes LLM ook niet beschikbaar ({llm_exc}). "
+                        "Check Hermes (:8642) of Instellingen → AI / LLM."
                     ),
                 }
         return {