refactor(harness)!: consolidate the Pydantic-AI harness + remove tracing handler (#431)

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: declan-scale

CHANGELOG.md

@@ -5,6 +5,7 @@
 ### ⚠ BREAKING CHANGES
 
 * **harness:** removed the deprecated bespoke LangGraph tracing handler `create_langgraph_tracing_handler` (and its `AgentexLangGraphTracingHandler` class) from the public `agentex.lib.adk` surface. Span tracing is now derived from the canonical `StreamTaskMessage*` stream by `UnifiedEmitter` — wrap your run in the harness `*Turn` and drive `UnifiedEmitter.yield_turn` / `auto_send_turn`. The `agentex init` templates were migrated accordingly.
+* **harness:** removed the deprecated bespoke Pydantic-AI tracing handler `create_pydantic_ai_tracing_handler` (and its `AgentexPydanticAITracingHandler` class) from the public `agentex.lib.adk` surface. Span tracing is now derived from the canonical `StreamTaskMessage*` stream by `UnifiedEmitter` — wrap your run in `PydanticAITurn` and drive `UnifiedEmitter.yield_turn` / `auto_send_turn`. The `agentex init` templates were migrated accordingly.
 * **harness:** each harness now exposes exactly `_<harness>_sync.py` + `_<harness>_turn.py` under `agentex.lib.adk._modules`. The OpenAI harness `OpenAITurn` and `convert_openai_to_agentex_events` moved to `agentex.lib.adk._modules._openai_turn` / `_openai_sync`; back-compat shims remain at `agentex.lib.adk.providers._modules.{openai_turn,sync_provider}` for one release. Public facade names (`stream_pydantic_ai_events`, `stream_langgraph_events`, `emit_langgraph_messages`, etc.) are unchanged.
 
 ### Features

src/agentex/lib/adk/__init__.py

@@ -11,10 +11,8 @@
     emit_langgraph_messages,
     convert_langgraph_to_agentex_events,
 )
-from agentex.lib.adk._modules._pydantic_ai_async import stream_pydantic_ai_events
+from agentex.lib.adk._modules._pydantic_ai_turn import PydanticAITurn, stream_pydantic_ai_events
 from agentex.lib.adk._modules._pydantic_ai_sync import convert_pydantic_ai_to_agentex_events
-from agentex.lib.adk._modules._pydantic_ai_tracing import create_pydantic_ai_tracing_handler
-from agentex.lib.adk._modules._pydantic_ai_turn import PydanticAITurn
 from agentex.lib.adk._modules._claude_code_sync import convert_claude_code_to_agentex_events
 from agentex.lib.adk._modules._claude_code_turn import (
     ClaudeCodeTurn,
@@ -75,7 +73,6 @@
     # Pydantic AI
     "stream_pydantic_ai_events",
     "convert_pydantic_ai_to_agentex_events",
-    "create_pydantic_ai_tracing_handler",
     "PydanticAITurn",
     # Claude Code
     "convert_claude_code_to_agentex_events",

src/agentex/lib/adk/_modules/_pydantic_ai_async.py

@@ -41,14 +41,9 @@ async def handle_message_send(params):
 
 import json
 import inspect
-from typing import TYPE_CHECKING, Any, Callable, AsyncIterator
+from typing import Any, Callable, AsyncIterator
 
 from pydantic_ai.run import AgentRunResultEvent
-
-if TYPE_CHECKING:
-    from agentex.lib.adk._modules._pydantic_ai_tracing import (
-        AgentexPydanticAITracingHandler,
-    )
 from pydantic_ai.messages import (
     TextPart,
     PartEndEvent,
@@ -124,7 +119,6 @@ def _tool_return_content(result: ToolReturnPart | Any) -> Any:
 
 async def convert_pydantic_ai_to_agentex_events(
     stream_response: AsyncIterator[Any],
-    tracing_handler: "AgentexPydanticAITracingHandler | None" = None,
     on_result: Callable[[AgentRunResultEvent], Any] | None = None,
 ) -> AsyncIterator[StreamTaskMessageStart | StreamTaskMessageDelta | StreamTaskMessageFull | StreamTaskMessageDone]:
     """Convert a Pydantic AI agent event stream into Agentex stream events.
@@ -148,11 +142,6 @@ async def convert_pydantic_ai_to_agentex_events(
         stream_response: The async iterator yielded by Pydantic AI's
             ``agent.run_stream_events(...)`` context manager (or a stream of
             ``AgentStreamEvent`` items received in an ``event_stream_handler``).
-        tracing_handler: Optional handler from
-            ``create_pydantic_ai_tracing_handler(...)``. When provided, each
-            tool call in the run is also recorded as an Agentex child span
-            beneath the handler's configured ``parent_span_id``. Streaming
-            behavior is unchanged when omitted.
         on_result: Optional callback invoked with the terminal
             ``AgentRunResultEvent`` when the run completes. Both sync and
             async callables are accepted. No ``StreamTaskMessage*`` events are
@@ -306,26 +295,6 @@ async def convert_pydantic_ai_to_agentex_events(
             if message_index is None:
                 continue
             yield StreamTaskMessageDone(type="done", index=message_index)
-            # Tool-call parts end with the model's full args known. Open a
-            # tracing child span for the tool execution now; close it when
-            # FunctionToolResultEvent arrives below.
-            if tracing_handler is not None and isinstance(event.part, ToolCallPart) and event.part.tool_call_id:
-                args: dict[str, Any] | str | None
-                raw_args = event.part.args
-                if isinstance(raw_args, dict):
-                    args = dict(raw_args)
-                elif isinstance(raw_args, str):
-                    try:
-                        args = json.loads(raw_args) if raw_args else {}
-                    except json.JSONDecodeError:
-                        args = {"_raw": raw_args}
-                else:
-                    args = {}
-                await tracing_handler.on_tool_start(
-                    tool_call_id=event.part.tool_call_id,
-                    tool_name=event.part.tool_name,
-                    arguments=args,
-                )
 
         elif isinstance(event, FunctionToolResultEvent):
             result = event.part
@@ -345,11 +314,6 @@ async def convert_pydantic_ai_to_agentex_events(
                     content=content_payload,
                 ),
             )
-            if tracing_handler is not None and tool_call_id:
-                await tracing_handler.on_tool_end(
-                    tool_call_id=tool_call_id,
-                    result=content_payload,
-                )
 
         elif isinstance(event, (FunctionToolCallEvent, FinalResultEvent, AgentRunResultEvent)):
             # Already covered by PartStart/PartDelta/PartEnd events above, or