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

[declan-scale]

Summary

Fifth slice of #425. Mirrors the LangGraph consolidation (#430) for the Pydantic-AI harness.

Breaking

• Removed create_pydantic_ai_tracing_handler (and its handler class). Span tracing is derived from the canonical stream by UnifiedEmitter.

Consolidation

• _pydantic_ai_async.py + _pydantic_ai_tracing.py collapsed into _pydantic_ai_sync.py and _pydantic_ai_turn.py (stream_pydantic_ai_events).
• adk/__init__.py re-wired (pydantic imports only).
• Removes the now-redundant test_pydantic_ai_sync_unified.py.

Test plan

• pytest pydantic sync/async/turn + harness + conformance — 95 passed
• import agentex.lib.adk clean; handler gone from __all__

Notes

Stacked on #430. Retarget to next after the chain merges.

🤖 Generated with Claude Code

Greptile Summary

This PR completes the Pydantic-AI harness consolidation by removing the deprecated create_pydantic_ai_tracing_handler / AgentexPydanticAITracingHandler and collapsing the old _pydantic_ai_async.py + _pydantic_ai_tracing.py modules into _pydantic_ai_turn.py. Span tracing is now derived automatically from the canonical event stream by UnifiedEmitter, matching the LangGraph consolidation in #430.

• Deleted files: _pydantic_ai_async.py, _pydantic_ai_tracing.py, and test_pydantic_ai_sync_unified.py — their functionality is absorbed into _pydantic_ai_turn.py and test_pydantic_ai_sync.py.
• Signature cleanups: tracing_handler parameter removed from PydanticAITurn.__init__, convert_pydantic_ai_to_agentex_events, and stream_pydantic_ai_events; all tracing is now opt-in via UnifiedEmitter(trace_id=...).
• Shared test fixture: Duplicate local _FakeTracing classes in the harness tests are replaced with a single FakeTracing imported from tests/lib/core/harness/_fakes.py.

Confidence Score: 5/5

This refactor is safe to merge — it removes deprecated code paths in favor of the unified emitter, and all new and existing tests pass.

The consolidation is a pure refactor: functionality moves from old modules to new ones without introducing new behavior. The breaking change is clearly documented in CHANGELOG.md, test coverage is comprehensive (passthrough equivalence, span derivation, and both sync/async harness paths), and the deleted code has direct replacements.

No files require special attention

Important Files Changed

Filename	Overview
src/agentex/lib/adk/_modules/_pydantic_ai_turn.py	Absorbs stream_pydantic_ai_events from the deleted _pydantic_ai_async.py; removes tracing_handler parameter from PydanticAITurn and strips the TYPE_CHECKING import for the deleted tracing module. Logic is unchanged.
src/agentex/lib/adk/_modules/_pydantic_ai_sync.py	Drops tracing_handler parameter and its on_tool_start/on_tool_end call-sites from convert_pydantic_ai_to_agentex_events; removes unused TYPE_CHECKING guard for the deleted tracing module. All existing streaming logic is preserved.
src/agentex/lib/adk/init.py	Re-wires pydantic imports: stream_pydantic_ai_events + PydanticAITurn now come from _pydantic_ai_turn (single import); create_pydantic_ai_tracing_handler removed from imports and all.
tests/lib/adk/test_pydantic_ai_sync.py	Removes TestTracingHandlerSync (now obsolete); adds TestUnifiedSyncPathPassthrough and TestUnifiedSyncPathSpanDerivation covering the new emitter-based span derivation path. Imports FakeTracing from the shared harness fakes module.
tests/lib/core/harness/test_harness_pydantic_ai_async.py	Replaces local _FakeTracing (2-tuple started) with shared FakeTracing (3-tuple started) from _fakes; adjusts one unpacking site accordingly. No logic changes to the tests themselves.
tests/lib/core/harness/test_harness_pydantic_ai_sync.py	Same as the async harness test — replaces local _FakeTracing with shared FakeTracing and unpacks the third (input) element at the one assertion site that inspects the tuple.

Sequence Diagram

%%{init: {'theme': 'neutral'}}%%
sequenceDiagram
    participant Caller
    participant stream_pydantic_ai_events
    participant PydanticAITurn
    participant convert_pydantic_ai_to_agentex_events
    participant UnifiedEmitter

    note over Caller,UnifiedEmitter: New consolidated path (this PR)

    Caller->>stream_pydantic_ai_events: stream, task_id
    stream_pydantic_ai_events->>PydanticAITurn: "PydanticAITurn(stream, model=None)"
    stream_pydantic_ai_events->>UnifiedEmitter: "UnifiedEmitter(task_id, trace_id=None, parent_span_id=None)"
    stream_pydantic_ai_events->>UnifiedEmitter: auto_send_turn(turn)
    UnifiedEmitter->>PydanticAITurn: turn.events
    PydanticAITurn->>convert_pydantic_ai_to_agentex_events: stream (no tracing_handler)
    convert_pydantic_ai_to_agentex_events-->>PydanticAITurn: "StreamTaskMessage* events"
    PydanticAITurn-->>UnifiedEmitter: "StreamTaskMessage* events"
    UnifiedEmitter-->>UnifiedEmitter: derive spans (if trace_id provided)
    UnifiedEmitter-->>stream_pydantic_ai_events: TurnResult
    stream_pydantic_ai_events-->>Caller: result.final_text

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
sequenceDiagram
    participant Caller
    participant stream_pydantic_ai_events
    participant PydanticAITurn
    participant convert_pydantic_ai_to_agentex_events
    participant UnifiedEmitter

    note over Caller,UnifiedEmitter: New consolidated path (this PR)

    Caller->>stream_pydantic_ai_events: stream, task_id
    stream_pydantic_ai_events->>PydanticAITurn: "PydanticAITurn(stream, model=None)"
    stream_pydantic_ai_events->>UnifiedEmitter: "UnifiedEmitter(task_id, trace_id=None, parent_span_id=None)"
    stream_pydantic_ai_events->>UnifiedEmitter: auto_send_turn(turn)
    UnifiedEmitter->>PydanticAITurn: turn.events
    PydanticAITurn->>convert_pydantic_ai_to_agentex_events: stream (no tracing_handler)
    convert_pydantic_ai_to_agentex_events-->>PydanticAITurn: "StreamTaskMessage* events"
    PydanticAITurn-->>UnifiedEmitter: "StreamTaskMessage* events"
    UnifiedEmitter-->>UnifiedEmitter: derive spans (if trace_id provided)
    UnifiedEmitter-->>stream_pydantic_ai_events: TurnResult
    stream_pydantic_ai_events-->>Caller: result.final_text

Loading

_{Reviews (10): Last reviewed commit: "docs(changelog): record Pydantic-AI trac..." | Re-trigger Greptile}

[declan-scale]

@greptile review

[declan-scale]

@greptile review

[declan-scale]

@greptile review