colibri: enable the GPU backend (COLI_CUDA=1) — fixes "exited before READY"

colibri's GPU tiering is OFF unless COLI_CUDA=1 (CUDA/HIP) / COLI_METAL=1 is set — it
is NOT implied by shipping a CUDA binary. We passed CUDA_EXPERT_GB without it, so the
engine aborted at startup: "CUDA_EXPERT_GB requires COLI_CUDA=1" -> exited before READY.

_build_env now defaults the backend on for cuda/hip (COLI_CUDA=1) and metal
(COLI_METAL=1) targets — coderai ships a CUDA build and routes colibri to the CUDA
engine — and only sets CUDA_EXPERT_GB when the CUDA backend is enabled. extra_env is
applied last so COLI_CUDA=0 can force a CPU run.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LoSpEthysqmseCc6Geizty

codai/__init__.py

@@ -16,7 +16,7 @@
 
 # Canonical product version for CoderAI — single source of truth. Both the API
 # metadata and the admin web UI read from here.
-__version__ = "0.1.72"
+__version__ = "0.1.73"
 
 # Configure the CUDA caching allocator BEFORE torch is imported anywhere.
 # expandable_segments lets the allocator return freed pages to the driver even

codai/api/colibri_worker.py

@@ -479,13 +479,39 @@ def resolve_service_key(cfg, model_dir: Optional[str] = None):
 
 
 def _build_env(cfg) -> tuple:
-    """Engine environment: CUDA_EXPERT_GB + free-form extra_env KEY=VALUE pairs."""
+    """Engine environment: enable the GPU backend + CUDA_EXPERT_GB + free-form extra_env.
+
+    colibri's GPU tiering is OFF unless explicitly enabled — ``COLI_CUDA=1`` (CUDA/HIP)
+    or ``COLI_METAL=1`` (Apple). It is NOT implied by shipping a CUDA build, and
+    ``CUDA_EXPERT_GB`` without ``COLI_CUDA=1`` aborts startup ("CUDA_EXPERT_GB requires
+    COLI_CUDA=1"). Since coderai routes colibri to the CUDA engine and ships a CUDA
+    binary, default the backend on (overridable via extra_env, e.g. COLI_CUDA=0)."""
     env = os.environ.copy()
     applied = {}
+
+    target = (getattr(cfg, "build_target", "auto") or "auto").strip().lower()
+    if target in ("", "auto"):
+        target = "cuda"    # coderai ships a CUDA build and routes colibri to the CUDA engine
+    cuda_on = False
+    if target in ("cuda", "hip"):
+        if "COLI_CUDA" not in env:
+            env["COLI_CUDA"] = "1"
+            applied["COLI_CUDA"] = "1"
+        cuda_on = env.get("COLI_CUDA") not in ("0", "", None)
+    elif target == "metal":
+        if "COLI_METAL" not in env:
+            env["COLI_METAL"] = "1"
+            applied["COLI_METAL"] = "1"
+
+    # CUDA_EXPERT_GB is only valid with the CUDA backend enabled — setting it otherwise
+    # aborts the engine, so gate it on COLI_CUDA being on.
     ceg = (getattr(cfg, "cuda_expert_gb", "") or "").strip()
-    if ceg and "CUDA_EXPERT_GB" not in env:
+    if ceg and cuda_on and "CUDA_EXPERT_GB" not in env:
         env["CUDA_EXPERT_GB"] = ceg
         applied["CUDA_EXPERT_GB"] = ceg
+
+    # Free-form passthrough LAST so an explicit pair (e.g. COLI_CUDA=0 to force CPU)
+    # overrides the defaults above.
     extra_env = (getattr(cfg, "extra_env", "") or "").strip()
     if extra_env:
         for tok in shlex.split(extra_env):