feat(routing): Rollen-Aliase 'debugger' und 'doctor' für Muse-Glimmer-30B verdrahtet

backend/models_catalog.json

@@ -3,7 +3,7 @@
   "version": "2026-07-03",
   "models": [
     {
-      "role": "fast", "name": "Muse-Glimmer-30B", "repo": "unsloth/Muse-Glimmer-30B-GGUF",
+      "role": "debugger", "name": "Muse-Glimmer-30B", "repo": "unsloth/Muse-Glimmer-30B-GGUF",
       "family": "muse", "generation": 1.0, "total_params_b": 30, "active_params_b": 30,
       "moe": false, "quant": "Q4_K_XL", "ctx": 65536, "tools": true, "vision": true
     },

deploy/llama-swap.config.yaml

@@ -34,10 +34,13 @@ models:
       context: 65536
   Muse-Glimmer-30B:
     # Meta Muse Glimmer 30B: Dichtes 30B-Agenten-Modell mit DFlash-Speculative-Drafting
-    # und 1.8B Perception Multimodal Projector.
+    # und 1.8B Perception Multimodal Projector. Ideal als Runtime-Debugger & Fehler-Diagnostiker.
     cmd: |
       llama-server -m /srv/models/Muse-Glimmer-30B-GGUF/Muse-Glimmer-30B-UD-Q4_K_XL.gguf --host 127.0.0.1 --port ${PORT} -c 65536 -ngl 999 -fa on --no-mmap --mmproj /srv/models/Muse-Glimmer-30B-GGUF/mmproj-kquant.gguf --jinja --spec-type draft-dflash --spec-draft-model /srv/models/Muse-Glimmer-30B-GGUF/dflash-kquant.gguf
     ttl: 600
+    aliases:
+    - debugger
+    - doctor
     capabilities:
       in: [text, image]
       out: [text]