Improve agent reliability and token efficiency

tests/test_ollama.py

@@ -0,0 +1,47 @@
+import unittest
+from unittest.mock import patch
+
+from assistant_bot.ollama import OllamaClient
+
+
+class OllamaClientTests(unittest.TestCase):
+    def test_chat_logs_token_and_timing_metrics(self) -> None:
+        client = OllamaClient("http://localhost:11434")
+        response = {
+            "message": {"content": "  готово  "},
+            "prompt_eval_count": 2004,
+            "eval_count": 12,
+            "prompt_eval_duration": 91_400_000,
+            "eval_duration": 25_000_000,
+            "total_duration": 130_000_000,
+            "load_duration": 4_000_000,
+        }
+
+        with (
+            patch.object(
+                client,
+                "_request_json",
+                return_value=response,
+            ) as request_json,
+            self.assertLogs(
+                "assistant_bot.ollama",
+                level="INFO",
+            ) as captured,
+        ):
+            answer = client._chat(
+                "qwen3.5:9b",
+                [{"role": "user", "content": "Привет"}],
+                json_mode=True,
+            )
+
+        self.assertEqual(answer, "готово")
+        payload = request_json.call_args.args[2]
+        self.assertEqual(payload["format"], "json")
+        self.assertIn("input_tokens=2004", captured.output[0])
+        self.assertIn("output_tokens=12", captured.output[0])
+        self.assertIn("total_tokens=2016", captured.output[0])
+        self.assertIn("prompt_eval_ms=91.4", captured.output[0])
+
+
+if __name__ == "__main__":
+    unittest.main()