Python: [BREAKING] added SerializationMixin and applied to contents, agents, chat client… (#1012)

* added SerializationMixin and applied to contents, agents, chat clients, removed AFBaseModel

* fix annotations type

* mypy fixes

* fix tests

* fix serializable subvalues and added large docstring

* updated indents in code block

* fixed exported urls

python/packages/lab/lightning/README.md

@@ -51,7 +51,7 @@ async def math_agent(task: TaskType, llm: LLM) -> float:
         MCPStdioTool(name="calculator", command="uvx", args=["mcp-server-calculator"]) as mcp_server,
         ChatAgent(
             chat_client=OpenAIChatClient(
-                ai_model_id=llm.model,
+                model_id=llm.model,
                 api_key="your-api-key",
                 base_url=llm.endpoint,
             ),

python/packages/lab/lightning/samples/train_math_agent.py

@@ -168,7 +168,7 @@ async def math_agent(task: MathProblem, llm: LLM) -> float:
         MCPStdioTool(name="calculator", command="uvx", args=["mcp-server-calculator"]) as mcp_server,
         ChatAgent(
             chat_client=OpenAIChatClient(
-                ai_model_id=llm.model,  # This is the model being trained
+                model_id=llm.model,  # This is the model being trained
                 api_key=os.getenv("OPENAI_API_KEY") or "dummy",  # Can be dummy when connecting to training LLM
                 base_url=llm.endpoint,  # vLLM server endpoint provided by agent-lightning
             ),

python/packages/lab/lightning/samples/train_tau2_agent.py

@@ -103,14 +103,14 @@ class Tau2Agent(LitAgent):
         assistant_chat_client = OpenAIChatClient(
             base_url=llm.endpoint,  # vLLM endpoint for the model being trained
             api_key=openai_api_key,
-            ai_model_id=llm.model,  # Model ID being trained
+            model_id=llm.model,  # Model ID being trained
         )
 
         # User simulator: uses a fixed, capable model for consistent simulation
         user_simulator_chat_client = OpenAIChatClient(
             base_url=openai_base_url,  # External API endpoint
             api_key=openai_api_key,
-            ai_model_id="gpt-4.1",  # Fixed model for user simulator
+            model_id="gpt-4.1",  # Fixed model for user simulator
         )
 
         try:

python/packages/lab/lightning/tests/test_lightning.py

@@ -18,10 +18,6 @@ from openai.types.chat import ChatCompletion, ChatCompletionMessage
 from openai.types.chat.chat_completion import Choice
 
 
-def test_import():
-    """Test that the module can be imported."""
-
-
 @pytest.fixture
 def workflow_two_agents():
     """Test a workflow with two OpenAI chat agents where first agent's result passes to second agent."""
@@ -111,14 +107,12 @@ def workflow_two_agents():
             yield workflow
 
 
-@pytest.mark.asyncio
 async def test_openai_workflow_two_agents(workflow_two_agents):
     events = await workflow_two_agents.run("Please analyze the quarterly sales data")
 
     assert "Based on the analysis 'Analyzed data shows trend upward', I recommend investing" in events.get_outputs()
 
 
-@pytest.mark.asyncio
 async def test_observability(workflow_two_agents):
     r"""Expected trace tree: