ToolRunner

A class for managing and executing tools from LLM requests.

This class handles tool execution for a set of tools that can be used by an AI agent in a conversation.

Bases: Enum

Enum defining various events that can occur during tool execution.

__init__(tools=None, registry=None)

Parameters:

Name	Type	Description	Default
`tools`	`list`	Initial list of tools to add.	`None`
`registry`	`dict`	Custom tool registry to use.	`None`

on_event(event: Event, callback)

Register a callback function for a specific event.

Parameters:

Name	Type	Description	Default
`event`	`Event`	The event to listen for.	required
`callback`	`callable`	The function to call when the event occurs.	required

on(tool, callback)

Register a callback function for a specific tool.

Parameters:

Name	Type	Description	Default
`tool`	`str`	The name of the tool.	required
`callback`	`callable`	The function to call when the tool is used.	required

add_tool(tool, agent=None)

Add a tool to the ToolRunner.

Parameters:

Name	Type	Description	Default
`tool`	`str`	The name of the tool to add.	required
`agent`	`object`	The agent object to attach the tool to.	`None`

add_tools(tools, agent=None)

Add multiple tools to the ToolRunner.

Parameters:

Name	Type	Description	Default
`tools`	`list`	A list of tool names to add.	required
`agent`	`object`	The agent object to attach the tools to.	`None`

remove_tool(tool)

Remove a tool from the ToolRunner.

Parameters:

Name	Type	Description	Default
`tool`	`str`	The name of the tool to remove.	required

loop(convo, llm, agent=None)

Start the main execution loop for processing conversations.

Parameters:

Name	Type	Description	Default
`convo`	`object`	The conversation object.	required
`llm`	`object`	The language model object.	required
`agent`	`object`	The agent object.	`None`

Returns:

Name	Type	Description
`tuple`		A tuple containing the final result and a list of actions performed.

execute(convo, llm, agent=None)

Execute a single step in the conversation processing.

Parameters:

Name	Type	Description	Default
`convo`	`object`	The conversation object.	required
`llm`	`object`	The language model object.	required
`agent`	`object`	The agent object.	`None`

Returns:

Name	Type	Description
`tuple`		A tuple containing the result and a log of the execution.

stop(result=None)

Stop the execution loop.

Parameters:

Name	Type	Description	Default
`result`	`any`	The final result to return.	`None`