test(daemon): ✅ Add test cases for task-based routing behavior in daemon

Co-Authored-By: Lilith Autocommit <noreply@atlilith.com>

tests/test_daemon.py

@@ -43,28 +43,55 @@ class TestMultiModelClientQueueParams:
 
     @pytest.mark.asyncio
     async def test_chat_passes_queue_params(self) -> None:
-        """_chat() sends keep_alive and context through to InferenceClient."""
+        """_chat() POSTs task + queue params to /v1/chat/completions."""
         from auto_commit_service.llm.multi_model_client import MultiModelLlamaClient
 
         client = MultiModelLlamaClient(client_id="test-svc")
         client._client = MagicMock()
-        client._client.chat = AsyncMock(return_value="response text")
+        client._client._coordinator_url = "http://coord.test"
+        client._client._default_priority = "normal"
 
+        captured: dict = {}
+
+        class _FakeResponse:
+            def raise_for_status(self) -> None:
+                return None
+
+            def json(self) -> dict:
+                return {"choices": [{"message": {"content": "response text"}}]}
+
+        class _FakeAsyncClient:
+            def __init__(self, *_args, **_kwargs) -> None:
+                pass
+
+            async def __aenter__(self):
+                return self
+
+            async def __aexit__(self, *_args) -> None:
+                return None
+
+            async def post(self, url: str, json: dict) -> _FakeResponse:
+                captured["url"] = url
+                captured["body"] = json
+                return _FakeResponse()
+
+        import httpx
+        with patch.object(httpx, "AsyncClient", _FakeAsyncClient):
             result = await client._chat(
-            "ministral-3b-instruct",
+                "summarization.short",
                 [{"role": "user", "content": "test"}],
                 max_tokens=100,
             )
 
         assert result == "response text"
-        client._client.chat.assert_called_once_with(
+        assert captured["url"] == "http://coord.test/v1/chat/completions"
-            model="ministral-3b-instruct",
+        assert captured["body"]["task"] == "summarization.short"
-            messages=[{"role": "user", "content": "test"}],
+        assert captured["body"]["messages"] == [{"role": "user", "content": "test"}]
-            max_tokens=100,
+        assert captured["body"]["max_tokens"] == 100
-            temperature=client._temperature,
+        assert captured["body"]["temperature"] == client._temperature
-            keep_alive=300,
+        assert captured["body"]["x_client_id"] == "test-svc"
-            context=None,
+        assert captured["body"]["x_keep_alive"] == 300
-        )
+        assert "model" not in captured["body"]
 
 
 class TestDaemonSimplifiedLoop: