feat: add configurable timeout for LLM requests

strix/interface/main.py

@@ -208,9 +208,12 @@ async def warm_up_llm() -> None:
             {"role": "user", "content": "Reply with just 'OK'."},
         ]
 
+        llm_timeout = int(os.getenv("LLM_TIMEOUT", "600"))
+
         response = litellm.completion(
             model=model_name,
             messages=test_messages,
+            timeout=llm_timeout,
         )
 
         validate_llm_response(response)

strix/llm/config.py

@@ -8,6 +8,7 @@ class LLMConfig:
         temperature: float = 0,
         enable_prompt_caching: bool = True,
         prompt_modules: list[str] | None = None,
+        timeout: int | None = None,
     ):
         self.model_name = model_name or os.getenv("STRIX_LLM", "openai/gpt-5")
 
@@ -17,3 +18,5 @@ class LLMConfig:
         self.temperature = max(0.0, min(1.0, temperature))
         self.enable_prompt_caching = enable_prompt_caching
         self.prompt_modules = prompt_modules or []
+
+        self.timeout = timeout or int(os.getenv("LLM_TIMEOUT", "600"))

strix/llm/llm.py

@@ -123,7 +123,10 @@ class LLM:
         self._total_stats = RequestStats()
         self._last_request_stats = RequestStats()
 
-        self.memory_compressor = MemoryCompressor()
+        self.memory_compressor = MemoryCompressor(
+            model_name=self.config.model_name,
+            timeout=self.config.timeout,
+        )
 
         if agent_name:
             prompt_dir = Path(__file__).parent.parent / "agents" / agent_name
@@ -359,7 +362,7 @@ class LLM:
             "model": self.config.model_name,
             "messages": messages,
             "temperature": self.config.temperature,
-            "timeout": 180,
+            "timeout": self.config.timeout,
         }
 
         if self._should_include_stop_param():

strix/llm/memory_compressor.py

@@ -85,6 +85,7 @@ def _extract_message_text(msg: dict[str, Any]) -> str:
 def _summarize_messages(
     messages: list[dict[str, Any]],
     model: str,
+    timeout: int = 600,
 ) -> dict[str, Any]:
     if not messages:
         empty_summary = "<context_summary message_count='0'>{text}</context_summary>"
@@ -106,7 +107,7 @@ def _summarize_messages(
         completion_args = {
             "model": model,
             "messages": [{"role": "user", "content": prompt}],
-            "timeout": 180,
+            "timeout": timeout,
         }
 
         response = litellm.completion(**completion_args)
@@ -146,9 +147,11 @@ class MemoryCompressor:
         self,
         max_images: int = 3,
         model_name: str | None = None,
+        timeout: int = 600,
     ):
         self.max_images = max_images
         self.model_name = model_name or os.getenv("STRIX_LLM", "openai/gpt-5")
+        self.timeout = timeout
 
         if not self.model_name:
             raise ValueError("STRIX_LLM environment variable must be set and not empty")
@@ -202,7 +205,7 @@ class MemoryCompressor:
         chunk_size = 10
         for i in range(0, len(old_msgs), chunk_size):
             chunk = old_msgs[i : i + chunk_size]
-            summary = _summarize_messages(chunk, model_name)
+            summary = _summarize_messages(chunk, model_name, self.timeout)
             if summary:
                 compressed.append(summary)
 

strix/tools/agents_graph/agents_graph_actions.py

@@ -230,9 +230,18 @@ def create_agent(
 
         state = AgentState(task=task, agent_name=name, parent_id=parent_id, max_iterations=300)
 
-        llm_config = LLMConfig(prompt_modules=module_list)
-
         parent_agent = _agent_instances.get(parent_id)
+
+        timeout = None
+        if (
+            parent_agent
+            and hasattr(parent_agent, "llm_config")
+            and hasattr(parent_agent.llm_config, "timeout")
+        ):
+            timeout = parent_agent.llm_config.timeout
+
+        llm_config = LLMConfig(prompt_modules=module_list, timeout=timeout)
+
         agent_config = {
             "llm_config": llm_config,
             "state": state,