Print text with color formatting. Args: text (str): The text to print color (str, optional): The color to use. Defaults to "

Push a message to the memory.

Animate a thinking spinner while a task is being executed. It use a daemon thread to run the animation. This will not block the main thread.

check if a bash command is unsafe. Tokens are basename-normalized so '/bin/rm' and './rm' still match the bare 'rm' entry. Multi-token e

Extract code/query blocks from LLM-generated text and process them for execution. This method parses the text looking for code blocks

Use MiniMax API to generate text via OpenAI-compatible interface. Supported models: - MiniMax-M3: Latest flagship model with

Clear all memory except system prompt

Use LiteLLM AI gateway for completion. Routes to 100+ providers (OpenAI, Anthropic, Azure, Bedrock, Vertex AI, Groq, Together

Asynchronously ask the LLM to process the prompt.

Parses agent tasks from the given LLM text. This method extracts task information from a JSON. It identifies task names and their det

Convert text to speech using an AI model and play the audio. Args: sentence (str): The text to convert to speech. Will b

Get a parameter name. Args: block (str): The block of text to search for the parameter parameter_name (str):

Check if a chromedriver binary is compatible with the installed Chrome version.

Check if an address is online by sending a ping request.

Extract form written by the LLM in format [input_name](value)

Extract all links from a sentence.

Get the major version of a chromedriver binary. Returns empty string on failure.

Select the first unvisited link that is not the current page. Preference is given to links not in search_history.

Ensure no '.' at the end of link

Compress (summarize) the memory using the model.

Create a Chrome WebDriver with specified options.

Create log dir

Detect the language of the given text using langdetect Limited to the supported languages list because of the model tendency to mista

Execute all the tools the agent has and return the result.

Check if executable blocks were found.

Get the filename for the save file.

Estimate context size based on the model name. EXPERIMENTAL for memory compression

Get the text content of the current page.

Navigate to a specified URL.

Load javascript from script folder to inject to page.

Load the memory from the last session.

Remove all code/query blocks within a tag from the answer text.

Save the session memory to a file.

Save the current session.

Create Chrome options - separated for reusability.

Abstract method that must be implemented by child classes to execute the tool's functionality. Args: blocks (List[str]):

Find all inputs elements on the page.

Generate text using the model. args: history: list of strings returns: None

Extract all input from the page and return them.

Get all navigable links on the current page.

Get a random user agent string with associated vendor.

Pull the next available recognition result from the audio queue. Args: last_spoken: Most recent TTS output, used to filt

Initialize STT and start listening to the microphone.

Return True if audio is currently being discarded.

Detect if code is running inside a Docker container.

Load the LLM router model. returns: AdaptiveClassifier: The loaded model exceptions: Exception: If th

Asks the LLM to make a plan. Args: prompt (str): The prompt to be sent to the LLM. max_retries (int): Maximum

Use the choosen provider to generate text.

Save code or query blocks to a file at the specified path. Creates the directory path if it doesn't exist. Args:

Take a screenshot of the current page, attempt to capture the full page by zooming out.

Select the appropriate agent based on the text. Args: text (str): The text to select the agent from Returns:

Show the answer in a pretty way. Show code blocks and their respective feedback by inserting them in the ressponse.

Displays the plan made by the agent. Args: agents_tasks (dict): The tasks assigned to each agent. answer (str

Speak the answer to the user in a non-blocking thread.

Start capturing audio from the microphone.

Summarize the text using the AI model. Args: text (str): The text to summarize min_length (int, optional): Th

Request AI agents to process the user input.

Translate the given text to English Args: text: string to translate origin_lang: ISO language code Re

Clamp a requested index into the valid voice range.

Check if any confirmation phrase ends the user's turn.

Return the last text spoken by TTS (for echo filtering).

Resolve the next index from a normalized browsing choice. Args: choice: 'next' or an explicit int index. current_idx:

Read and normalize the user's voice-browsing input. Args: current_idx: Index of the voice just auditioned. voice_coun