fix(llm): support flexible base_url formats for open-ai compatible proxies like orcarouter

core/llm.py

@@ -16,17 +16,17 @@ class LLMClient:
         model: Optional[str] = None,
         base_url: Optional[str] = None,
     ):
-        self.provider = provider or os.getenv("AI_PROVIDER", "gemini").lower()
+        self.provider = provider or os.getenv("AI_PROVIDER", "openai").lower()
         self.api_key = api_key or os.getenv("AI_API_KEY", "")
-        self.model = model or os.getenv("AI_MODEL", "gemini-1.5-flash" if self.provider == "gemini" else "gpt-4o-mini")
-        self.base_url = base_url or os.getenv("AI_BASE_URL")
+        self.model = model or os.getenv("AI_MODEL", "orcarouter/auto" if self.provider == "openai" else "gemini-1.5-flash")
+        self.base_url = base_url or os.getenv("AI_BASE_URL", "https://api.orcarouter.ai/v1")
 
     async def generate_json(self, prompt: str, system_prompt: Optional[str] = None, action_name: str = "general") -> Dict[str, Any]:
         """Send prompt to LLM and parse JSON response."""
         start_time = time.time()
         status = "error"
         try:
-            if self.provider == "gemini":
+            if self.provider == "gemini" and "orcarouter" not in (self.base_url or ""):
                 result = await self._call_gemini(prompt, system_prompt)
             else:
                 result = await self._call_openai(prompt, system_prompt)
@@ -66,7 +66,9 @@ class LLMClient:
             return json.loads(raw_text)
 
     async def _call_openai(self, prompt: str, system_prompt: Optional[str] = None) -> Dict[str, Any]:
-        url = self.base_url or "https://api.openai.com/v1/chat/completions"
+        base = (self.base_url or "https://api.orcarouter.ai/v1").rstrip("/")
+        url = base if base.endswith("/chat/completions") else f"{base}/chat/completions"
+
         headers = {
             "Authorization": f"Bearer {self.api_key}",
             "Content-Type": "application/json"