Execute the task with maximum number of steps

Custom serialization handling circular references

Safely get an accessibility attribute with error reporting

Reset all state variables

Get accumulated terminal output

Add browser state as human message

Performs a specified accessibility action on an element.

Build UI tree for a specific application

Cleanup observers and release resources

Simulates a click on a Mac UI element.

Convert the UI tree to a string representation focusing on interactive and context elements

Get the list of agents for an automation flow

Initialize LLM based on provider

Get saved API key from .env file based on provider

Get all actions from history

Save API key to .env file based on provider

Save user preferences to JSON file

Simulates typing text into a Mac UI element with action-based submission

Get all errors from history

Extract the result text from the output by checking various formats.

Get the last pid from the last result

Check if the agent is done

Execute multiple actions

Remove last message from history

Update LLM provider and model preferences

Enhanced string representation including more attributes

Representation of the AgentHistoryList object

Convert dictionary-based actions to ActionModel instances

Creates a Pydantic model from function signature

Should not be thread blocking because posthog magically handles it

Get available actions for an element with proper error handling

Get or launch target application PID

Handle pause and stop flags. Returns True if execution should continue.

Handle all types of errors that can occur during a step

Handle UI change notifications

Determine if an element is truly interactive based on its role and actions.

Log the agent run

Log the model's response

Create and store history item

Process a single UI element

Register all default browser actions

Save conversation history to file if path is specified

Setup dynamic action models from controller's registry

Set up accessibility observer for UI changes

Setup accessibility observer for an application

Check if we should stop due to too many failures

Write messages to conversation file

Write model response to conversation file

Execute an action

Get all results from history

Comprehensively adds a new logging level to the `logging` module and the currently configured logging class. `levelName` becomes an attribute of

Add a new agent to an automation flow

Add a new automation flow

Add a message with metadata

Click element by index

Clean up application session

Creates a Pydantic model from registered actions

Create the Gradio interface with all components.

Execute a registered action

Find an available port starting from start_port

Focus a specific application window

Format error message based on error type and optionally include trace

Return a detailed string with all attributes of the element

Get name of frontmost application

Get a direct response from the LLM model without using the Agent

Get next action from LLM based on current state

Get a description of all actions for the prompt

Get current application UI state

Creates and returns a SystemMessage with formatted content.

Creates and returns a HumanMessage with formatted content.

Returns a string containing important rules for the system.

Initialize application session

Returns a string describing the expected input format.

Load user preferences from JSON file