Stability: CPU fix with psutil interval and LLM timeouts

ai_agent.py

@@ -11,21 +11,30 @@ def get_llm_response(prompt: str, provider: str, cfg: dict) -> str:
         api_key = cfg.get("gemini_api_key") or os.getenv("GEMINI_API_KEY")
         url = f"https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent?key={api_key}"
         payload = {"contents": [{"parts": [{"text": prompt}]}]}
-        res = requests.post(url, json=payload)
-        if res.status_code == 200:
-            return res.json()["candidates"][0]["content"]["parts"][0]["text"]
-        return f"Erro Gemini: {res.text}"
+        try:
+            res = requests.post(url, json=payload, timeout=30)
+            if res.status_code == 200:
+                data = res.json()
+                if "candidates" in data and len(data["candidates"]) > 0:
+                     return data["candidates"][0]["content"]["parts"][0]["text"]
+                return f"Erro Gemini (Dados Vazios): {res.text}"
+            return f"Erro Gemini (Status {res.status_code}): {res.text}"
+        except Exception as e:
+            return f"Erro de Conexão Gemini: {str(e)}"
     
     elif provider == "ollama":
         ollama_host = os.getenv("OLLAMA_HOST", "http://ollama-lw4s8g4gc8gss4gkc4gg0wk4:11434")
-        res = requests.post(f"{ollama_host}/api/generate", json={
-            "model": os.getenv("OLLAMA_MODEL", "qwen2.5-coder:1.5b"),
-            "prompt": prompt,
-            "stream": False
-        })
-        if res.status_code == 200:
-            return res.json().get("response", "")
-        return f"Erro Ollama: {res.text}"
+        try:
+            res = requests.post(f"{ollama_host}/api/generate", json={
+                "model": os.getenv("OLLAMA_MODEL", "qwen2.5-coder:1.5b"),
+                "prompt": prompt,
+                "stream": False
+            }, timeout=30)
+            if res.status_code == 200:
+                return res.json().get("response", "")
+            return f"Erro Ollama (Status {res.status_code}): {res.text}"
+        except Exception as e:
+            return f"Erro de Conexão Ollama: {str(e)}"
     
     return "Provedor desconhecido."
 
@@ -92,6 +101,8 @@ Resposta: A memória RAM está operando com 20% de uso.
     print(f"--- INICIANDO AGENTE ({provider}) ---")
     
     for i in range(max_loops):
+        import time
+        time.sleep(0.5) # Respiro para a CPU
         print(f"\n[LOOP {i+1}/{max_loops}]")
         full_prompt = system_prompt + current_iteration_history
         response = get_llm_response(full_prompt, provider, cfg)