Increase Ollama timeout to 180s and add num_ctx

ai_agent.py

@@ -24,12 +24,14 @@ def get_llm_response(prompt: str, provider: str, cfg: dict) -> str:
     
     elif provider == "ollama":
         ollama_host = os.getenv("OLLAMA_HOST", "http://ollama:11434")
+        model = os.getenv("OLLAMA_MODEL", "qwen2.5-coder:1.5b")
         try:
             res = requests.post(f"{ollama_host}/api/generate", json={
-                "model": os.getenv("OLLAMA_MODEL", "qwen2.5-coder:1.5b"),
+                "model": model,
                 "prompt": prompt,
-                "stream": False
-            }, timeout=30)
+                "stream": False,
+                "options": {"num_ctx": 4096}
+            }, timeout=180)
             if res.status_code == 200:
                 return res.json().get("response", "")
             return f"Erro Ollama (Status {res.status_code}): {res.text}"