Solving Captchas with Amazon Nova Act: A Developer’s Guide

mkdir nova && cd nova
python -m venv .venv
source .venv/bin/activate
pip install nova_act

import fire  # type: ignore

from nova_act import NovaAct

def main(user_data_dir: str = None, headless: bool = False) -> None:
    """
    Automates interaction with the MTCaptcha demo page using NovaAct.

    Args:
        user_data_dir (str, optional): Directory to store user data (e.g., for Chrome profile). Defaults to None.
        headless (bool, optional): Whether to run the browser in headless mode. Defaults to False.
    """
    with NovaAct(
        starting_page="https://2captcha.com/demo/mtcaptcha",
        user_data_dir=user_data_dir,
        headless=headless,
    ) as nova:
        nova.act(
            "If a cookie consent banner appears, close it. "
            "View the case-sensitive captcha character text at the center of the page. "
            "Enter the captcha text in the input field with placeholder 'Enter text from image'. "
            "Click the Check button."
        )

if __name__ == "__main__":
    fire.Fire(main)


Administrator@EC2AMAZ-LHMLS5H MINGW64 ~/Desktop/Nova/nova-act/src/nova_act/samples (main)
$ python captcha.py 
2025-04-29 07:33:54,840 - nova_act.impl.playwright - WARNING - The Nova Act SDK is unable to run with `chrome_channel='chrome'` and is falling back to 'chromium'. If you wish to use an alternate `chrome_channel`, please install it with `python -m playwright install chrome`. For more information, please consult Playwright's documentation: https://playwright.dev/python/docs/browsers.

start session 1489e281-f900-46ba-8616-0c0ac9627663 on https://2captcha.com/demo/mtcaptcha logs dir C:\Users\ADMINI~1\AppData\Local\Temp\2\tmpju27dmhq_nova_act_logs\1489e281-f900-46ba-8616-0c0ac9627663  

1489> act("If a cookie consent banner appears, close it. View the case-sensitive captcha characters  text at center of page. Enter the captcha text with in the input field with placeholder 'Enter text from image'. Click Check button. ")
............
1489> think("I am on the 2captcha website. My task is to view the case-sensitive captcha characters text at the center of the page, enter the captcha text within the input field with placeholder 'Enter text from image', and click the Check button. The captcha is visible at the center of the page, and the text is 'jlljc'. The input field is empty, as indicated by the placeholder text 'Enter text from image'. I should type 'jlljc' into the input field to populate it.");
>> agentType("jlljc", "<box>388,504,428,762</box>");
..........
1489> think("The input field now contains 'jlljc', so my last action was successful. I have not checked the captcha yet, as I see the Check button is still visible. I should click the Check button to check the captcha.");
>> agentClick("<box>500,518,540,640</box>");
..............
1489> think("The captcha has been checked, as indicated by the 'Captcha is passed successfully!' message. Therefore, my last action was a success. I have viewed the case-sensitive captcha characters text, entered the captcha text in the input field, and clicked the Check button. The task is complete, and I was not asked to provide specific information, so I will simply return.");
>> return;
.
end session