Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Comet MCP – Give Claude Code a browser that can click

github.com

Post reply on HN
Print view View original

21–29 of 29 posts

Re: Show HN: Comet MCP – Give Claude Code a browser that can click

#23

Post by rvz »

observationist wrote: post #14
DougN7 wrote: post #11

I didn’t realize AI could interact with browsers like this already (guess I’m naive). Isn’t this setting up for the scenario where the AI is duped into logging into your bank account and transferring your money away? Not sure I have enough trust to allow an AI to touch a browser.

People are already going full Leroy Jenkins with this stuff, and OpenAI, other labs are snarfing up their usage data. Hopefully with their brave sacrifice, they can figure out all the security pitfalls before it becomes common enough that someone with a clever jailbreak ends up pulling of a billion dollar heist, or orders pizza for half the country. It's 100% absolutely not safe yet. You can effectively copy and past…

Great time to be an offensive security researcher specialising in researching LLM adversarial attacks.

Re: Show HN: Comet MCP – Give Claude Code a browser that can click

#24

Post by observationist »

rvz wrote: post #23
observationist wrote: post #14

Earlier quoted context omitted.

People are already going full Leroy Jenkins with this stuff, and OpenAI, other labs are snarfing up their usage data. Hopefully with their brave sacrifice, they can figure out all the security pitfalls before it becomes common enough that someone with a clever jailbreak ends up pulling of a billion dollar heist, or orders pizza for half the country. It's 100% absolutely not safe yet. You can effectively copy and past…

Great time to be an offensive security researcher specialising in researching LLM adversarial attacks.

Yeah - the red team folks probably have one of the most fun jobs in the world right now.

Re: Show HN: Comet MCP – Give Claude Code a browser that can click

#25

Post by tbocek »

DougN7 wrote: post #11

I didn’t realize AI could interact with browsers like this already (guess I’m naive). Isn’t this setting up for the scenario where the AI is duped into logging into your bank account and transferring your money away? Not sure I have enough trust to allow an AI to touch a browser.

That's a valid concern. I took a more constrained approach for web searches for exactly this reason. Instead of giving the LLM full browser control, I built a Firefox extension that only handles web search client-side.

When my local LLM (llama.cpp) needs to search, it opens DuckDuckGo in a new window, loads the result pages in tabs, extracts content with Readability.js, and feeds it back. You stay in the loop - can see what's loading, solve captchas if needed. Less autonomous than Comet/Playwright, with a narrower use-case, but also less risk.

Its still a prototype though: https://github.com/tbocek/llm-local-web-search

Re: Show HN: Comet MCP – Give Claude Code a browser that can click

#27

Post by luckman212 »

observationist wrote: post #24
rvz wrote: post #23

Earlier quoted context omitted.

Great time to be an offensive security researcher specialising in researching LLM adversarial attacks.

Yeah - the red team folks probably have one of the most fun jobs in the world right now.

Depends on your definition of "fun"
Post reply on HN