search

Skyvern-AI / skyvern

Automate browser-based workflows with LLMs and Computer Vision
https://www.skyvern.com
GNU Affero General Public License v3.0
5.65k stars 405 forks source link

"--disable-blink-features=AutomationControlled" Really? #91

Closed RonaldRuckus closed 4 months ago

RonaldRuckus commented 5 months ago

I can understand wanting to use an LLM to browse the web.

Purposely obfuscating the fact that you're an automation tool is gross and supports people using more intrusive DRM to prevent bots from accessing their sites - at the expense of real visitors.

By all means this is automation software and should respect anti-bot protections.

For context:

"--disable-blink-features=AutomationControlled" is a command line argument that prevents Chromium from indicating that it's currently being controlled by automation software. This is a typical method used by data scrapers to obfuscate their program, bypassing anti-scraping/anti-bot protection.

For the good health of the WWW this service should be more respectful. Otherwise more intrusive methods will be introduced/implemented such as the intrusive Web Environment Integrity

suchintan commented 5 months ago

Thanks for opening this issue! We just added it because a lot of websites just dont work without it enabled

suchintan commented 5 months ago

Does anyone else have a strong opinion either way? We will let the community decide whether they want to remove it or keep it. Please 👍 or 👎 !

Keep it = 👍 Remove it = 👎

RonaldRuckus commented 5 months ago

I feel like this is like asking car drivers if they want to add or remove pedestrian walking strips.

Implementing this is a push towards more intrusive anti-bot security that will ultimately be the demise of this project and closed-off WWW. If a website doesn't want automated sessions it should be respected.

suchintan commented 5 months ago

The community is pretty split here. I'll leave this issue open in case anyone else has an opinion

FWIW -- the reason we left this in was because this is a pretty well understood chrome functionality -- it just goes one step further than the stuff mentioned here (https://stackoverflow.com/questions/43143014/chrome-is-being-controlled-by-automated-test-software)