Fix: Propagate timeout to AI backend and increase default to 60s

src/nacc_orchestrator/agents.py

@@ -21,7 +21,7 @@ class AgentBackendError(RuntimeError):
 
 
 class LLMBackend:
-    def complete(self, prompt: str, *, context: dict[str, Any] | None = None) -> str:  # pragma: no cover - interface
+    def complete(self, prompt: str, *, context: dict[str, Any] | None = None, timeout: float | None = None) -> str:  # pragma: no cover - interface
         raise NotImplementedError
 
 
@@ -40,7 +40,7 @@ class DockerMistralBackend(LLMBackend):
         self.timeout = timeout
         self.environment = environment
 
-    def complete(self, prompt: str, *, context: dict[str, Any] | None = None) -> str:
+    def complete(self, prompt: str, *, context: dict[str, Any] | None = None, timeout: float | None = None) -> str:
         # Merge prompt with context
         merged_prompt = prompt
         if context:
@@ -54,7 +54,8 @@ class DockerMistralBackend(LLMBackend):
             ["docker", "model", "run", self.model_name, merged_prompt],
             text=True,
             capture_output=True,
-            timeout=self.timeout,
+            capture_output=True,
+            timeout=timeout or self.timeout,
             env=env,
         )
         
@@ -68,7 +69,7 @@ class DockerMistralBackend(LLMBackend):
 class LocalHeuristicBackend(LLMBackend):
     """Deterministic fallback backend for development/testing."""
 
-    def complete(self, prompt: str, *, context: dict[str, Any] | None = None) -> str:
+    def complete(self, prompt: str, *, context: dict[str, Any] | None = None, timeout: float | None = None) -> str:
         summary = {
             "prompt_hash": hash(prompt) & 0xFFFFFFFF,
             "context_keys": sorted(list((context or {}).keys())),
@@ -273,9 +274,9 @@ class AgentSuite:
         request = RouterRequest(task=description, required_tags=preferred_tags, parallelism=1)
         return self.router.select_nodes(request)
 
-    def probe_backend(self, message: str = "NACC orchestrator health check", context: dict[str, Any] | None = None) -> str:
+    def probe_backend(self, message: str = "NACC orchestrator health check", context: dict[str, Any] | None = None, timeout: float | None = None) -> str:
         probe_context = context or {"source": "nacc-orchestrator", "kind": "health-check"}
-        return self.backend.complete(message, context=probe_context)
+        return self.backend.complete(message, context=probe_context, timeout=timeout)
 
 
 __all__ = [

src/nacc_orchestrator/blaxel_backend.py

@@ -72,7 +72,7 @@ class BlaxelBackend:
             # The first part uses the workload name (gemini-2-0-flash-exp), second part uses actual model
             self.endpoint_url = f"{base_url}/{workspace}/models/gemini-2-0-flash-exp/v1beta/models/{model}:generateContent"
     
-    def complete(self, prompt: str, *, context: dict[str, Any] | None = None) -> str:
+    def complete(self, prompt: str, *, context: dict[str, Any] | None = None, timeout: float | None = None) -> str:
         """
         Generate completion using Blaxel Model Gateway.
         
@@ -132,7 +132,8 @@ class BlaxelBackend:
                 self.endpoint_url,
                 headers=headers,
                 json=payload,
-                timeout=self.timeout
+                json=payload,
+                timeout=timeout or self.timeout
             )
             response.raise_for_status()
             
@@ -177,7 +178,8 @@ class BlaxelBackend:
                 self.endpoint_url,
                 headers=headers,
                 json=payload,
-                timeout=self.timeout
+                json=payload,
+                timeout=timeout or self.timeout
             )
             response.raise_for_status()
             

src/nacc_orchestrator/server.py

@@ -46,7 +46,7 @@ class ChatPayload(BaseModel):
     session_id: str = Field(default="default")
     current_node: str | None = None
     current_path: str = Field(default="/home")
-    timeout: float = Field(default=30.0, gt=0.0, le=600.0)
+    timeout: float = Field(default=60.0, gt=0.0, le=600.0)
 
 
 def build_service(config_path: str | None = None) -> OrchestratorService:
@@ -276,7 +276,7 @@ def create_app(service: OrchestratorService) -> FastAPI:
             If the query doesn't match any tool, set "tool" to "general_response" and include your response in "reasoning"."""
             
             # Get AI decision
-            ai_result = service.check_agent_backend(ai_prompt, context)
+            ai_result = service.check_agent_backend(ai_prompt, context, timeout=payload.timeout)
             ai_response = ai_result.get("response", "")
             
             # Try to parse AI's structured response