v4 Schritt 3: 'Verbinden'-Tab (Orchestrierung)

- routers/integration.py: GET /api/integration/test ruft Engine /v1/models -> ok + Modell-Liste.
- connect.js: Verbindungstest-Button, Bild/Vision-Flow (erkennt Vision-Modell, sonst Cookbook-Hinweis,
  OpenAI-Vision-Snippet), Tool-Configs (OpenCode/Cline/Cursor/OpenWebUI) mit Copy.
- Nav-Eintrag 'Verbinden' + View + main.js Registrierung.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

routers/integration.py

@@ -0,0 +1,24 @@
+"""
+Integrations-Router: testet die OpenAI-kompatible Verbindung zur LLM-Engine.
+Damit der „Verbinden"-Assistent dem Nutzer sagen kann: funktioniert ✓ + welche Modelle.
+"""
+
+from fastapi import APIRouter, Depends
+
+from auth import auth
+from config import LLAMA_SWAP_URL
+from llamaswap import _swap_get
+
+router = APIRouter(prefix="/api/integration", dependencies=[Depends(auth)])
+
+
+@router.get("/test")
+def test_connection():
+    """Ruft die OpenAI-kompatible Modell-Liste der Engine ab (das, was externe Tools nutzen)."""
+    try:
+        data = _swap_get("/v1/models")
+        items = data.get("data", data) if isinstance(data, dict) else data
+        models = [m.get("id") for m in (items or []) if isinstance(m, dict) and m.get("id")]
+        return {"ok": True, "models": models, "url": LLAMA_SWAP_URL}
+    except Exception as exc:  # noqa: BLE001
+        return {"ok": False, "error": str(exc), "url": LLAMA_SWAP_URL}