You can subclass`WrapperToolset` to change the wrapped toolset's tool execution behavior by overriding the[`call_tool()`][pydantic_ai.toolsets.AbstractToolset.call_tool] method.

```python {title="logging_toolset.py" requires="function_toolset.py,combined_toolset.py,renamed_toolset.py,prepared_toolset.py"}

import asyncio

from typing_extensionsimport Any

from prepared_toolsetimport prepared_toolset

asyncdefcall_tool(self,name:str,tool_args: dict[str, Any],ctx: RunContext,tool: ToolsetTool) -> Any:

LOG.append(f'Calling tool{name!r} with args:{tool_args!r}')

try:

await asyncio.sleep(0.1*len(LOG))# (1)!

result=awaitsuper().call_tool(name, tool_args, ctx, tool)

LOG.append(f'Finished calling tool{name!r} with result:{result!r}')

exceptExceptionas e:

logging_toolset= LoggingToolset(prepared_toolset)

agent= Agent(TestModel(),toolsets=[logging_toolset])# (1)!

agent= Agent(TestModel(),toolsets=[logging_toolset])# (2)!

result= agent.run_sync('Call all the tools')

print(LOG)

1. We use[`TestModel`][pydantic_ai.models.test.TestModel] here as it will automatically call each tool.

1. All docs examples are tested in CI and their their output is verified, so we need`LOG` to always have the same order whenever this code is run. Since the tools could finish in any order, we sleep an increasing amount of time based on which number tool call we are to ensure that they finish (and log) in the same order they were called in.

2. We use[`TestModel`][pydantic_ai.models.test.TestModel] here as it will automatically call each tool.

_(This example is complete, it can be run "as is")_

_output_validators:list[OutputValidator[AgentDepsT,OutputDataT]]

_run_ctx:RunContext[AgentDepsT]

_usage_limits:UsageLimits|None

_toolset:ToolManager[AgentDepsT]

_tool_manager:ToolManager[AgentDepsT]

_agent_stream_iterator:AsyncIterator[AgentStreamEvent]|None=field(default=None,init=False)

_final_result_event:FinalResultEvent|None=field(default=None,init=False)

raiseexceptions.UnexpectedModelBehavior(# pragma: no cover

)

returnawaitself._toolset.handle_call(tool_call,allow_partial=allow_partial)

elifdeferred_tool_calls:=self._toolset.get_deferred_tool_calls(message.parts):

returnawaitself._tool_manager.handle_call(tool_call,allow_partial=allow_partial)

elifdeferred_tool_calls:=self._tool_manager.get_deferred_tool_calls(message.parts):

ifnotself._output_schema.allows_deferred_tool_calls:

raiseexceptions.UserError(# pragma: no cover

):# pragma: no branch

return_messages.FinalResultEvent(tool_name=None,tool_call_id=None)

elifisinstance(new_part,_messages.ToolCallPart)and (

tool_def:=self._toolset.get_tool_def(new_part.tool_name)

tool_def:=self._tool_manager.get_tool_def(new_part.tool_name)

):

iftool_def.kind=='output':

return_messages.FinalResultEvent(

_output_validators:list[OutputValidator[AgentDepsT,OutputDataT]]

_output_tool_name:str|None

_on_complete:Callable[[],Awaitable[None]]

_toolset:ToolManager[AgentDepsT]

_tool_manager:ToolManager[AgentDepsT]

_initial_run_ctx_usage:Usage=field(init=False)

is_complete:bool=field(default=False,init=False)

raiseexceptions.UnexpectedModelBehavior(# pragma: no cover

)

returnawaitself._toolset.handle_call(tool_call,allow_partial=allow_partial)

elifdeferred_tool_calls:=self._toolset.get_deferred_tool_calls(message.parts):

returnawaitself._tool_manager.handle_call(tool_call,allow_partial=allow_partial)

elifdeferred_tool_calls:=self._tool_manager.get_deferred_tool_calls(message.parts):

ifnotself._output_schema.allows_deferred_tool_calls:

raiseexceptions.UserError(