[claude] Document and validate AirLLM Apple Silicon requirements (#1284) (#1298)

src/config.py

@@ -94,8 +94,9 @@ class Settings(BaseSettings):
 
     # ── Backend selection ────────────────────────────────────────────────────
     # "ollama"  — always use Ollama (default, safe everywhere)
+    # "airllm"  — AirLLM layer-by-layer loading (Apple Silicon only; degrades to Ollama)
     # "auto"    — pick best available local backend, fall back to Ollama
-    timmy_model_backend: Literal["ollama", "grok", "claude", "auto"] = "ollama"
+    timmy_model_backend: Literal["ollama", "airllm", "grok", "claude", "auto"] = "ollama"
 
     # ── Grok (xAI) — opt-in premium cloud backend ────────────────────────
     # Grok is a premium augmentation layer — local-first ethos preserved.

src/timmy/agent.py

@@ -301,6 +301,26 @@ def create_timmy(
 
         return GrokBackend()
 
+    if resolved == "airllm":
+        # AirLLM requires Apple Silicon.  On any other platform (Intel Mac, Linux,
+        # Windows) or when the package is not installed, degrade silently to Ollama.
+        from timmy.backends import is_apple_silicon
+
+        if not is_apple_silicon():
+            logger.warning(
+                "TIMMY_MODEL_BACKEND=airllm requested but not running on Apple Silicon "
+                "— falling back to Ollama"
+            )
+        else:
+            try:
+                import airllm  # noqa: F401
+            except ImportError:
+                logger.warning(
+                    "AirLLM not installed — falling back to Ollama. "
+                    "Install with: pip install 'airllm[mlx]'"
+                )
+        # Fall through to Ollama in all cases (AirLLM integration is scaffolded)
+
     # Default: Ollama via Agno.
     model_name, is_fallback = _resolve_model_with_fallback(
         requested_model=None,