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

Show HN: A MitM proxy to see what your LLM tools are sending

github.com

Post reply on HN
Print view View original

101–110 of 128 posts

Re: Show HN: A MitM proxy to see what your LLM tools are sending

#101

Post by 101008 »

catlifeonmars wrote: post #54

This tool looks like it unconditionally disables tls verification for upstream requests. It shells out to mitmproxy with "--set", "ssl_insecure=true" This took all of 5 minutes to find reading through main.py on my phone. https://github.com/jmuncor/sherlock/blob/fb76605fabbda351828... Edit: In case it’s not clear, you should not use this.

I think the main problem is when OP wrote:

> I built this

Instead of

> I prompted this

I am OK with people publishing new ideas to the web as long as they are 100% honest and admit they just had an idea and asked an AI to build it for themselves. That way I can understand they may not have considered all the things that needs to be considered and I can skip it (and then prompt it myself if I want to, adding all the things I consider necesary)

Re: Show HN: A MitM proxy to see what your LLM tools are sending

#102

Post by arowthway »

catlifeonmars wrote: post #54

This tool looks like it unconditionally disables tls verification for upstream requests. It shells out to mitmproxy with "--set", "ssl_insecure=true" This took all of 5 minutes to find reading through main.py on my phone. https://github.com/jmuncor/sherlock/blob/fb76605fabbda351828... Edit: In case it’s not clear, you should not use this.

Don't use it if you plan to auto accept terminal commands, without a sandbox, while on a public wifi in a cafe, next to a hacker who decides to bet on you running a very niche configuration.

Re: Show HN: A MitM proxy to see what your LLM tools are sending

#103

Post by catlifeonmars »

arowthway wrote: post #102
catlifeonmars wrote: post #54

This tool looks like it unconditionally disables tls verification for upstream requests. It shells out to mitmproxy with "--set", "ssl_insecure=true" This took all of 5 minutes to find reading through main.py on my phone. https://github.com/jmuncor/sherlock/blob/fb76605fabbda351828... Edit: In case it’s not clear, you should not use this.

Don't use it if you plan to auto accept terminal commands, without a sandbox, while on a public wifi in a cafe, next to a hacker who decides to bet on you running a very niche configuration.

All you need is to manipulate DNS, inject a record with a long TTL and the rest is not required.

It scales very well and I guarantee this is not the only instance of misconfigured host verification. In other words, this is not as niche as you might think.

Re: Show HN: A MitM proxy to see what your LLM tools are sending

#104

Post by monkaiju »

catlifeonmars wrote: post #54

This tool looks like it unconditionally disables tls verification for upstream requests. It shells out to mitmproxy with "--set", "ssl_insecure=true" This took all of 5 minutes to find reading through main.py on my phone. https://github.com/jmuncor/sherlock/blob/fb76605fabbda351828... Edit: In case it’s not clear, you should not use this.

And it's already surpassed my most starred project when it was on GitHub, all the more validating to have moved it to forgejo. If vibecoded stuff with unbelievable security vulns can get so much praise the whole star system doesn't work as a quality filter. Similarly a well crafted README used to help reflect quality, no longer...

Re: Show HN: A MitM proxy to see what your LLM tools are sending

#105

Post by catlifeonmars »

Balinares wrote: post #95
Lucasoato wrote: post #91

Earlier quoted context omitted.

Co-Authored-By: Claude Opus 4.5

Ah, getting the job done by disabling important validation, if that isn't the most prominent Opus trait... I wonder how much this will end up costing the industry in aggregate.

I’m thinking of pivoting into cybersecurity. I suspect that’s where the all money will be in the next couple of years.

Re: Show HN: A MitM proxy to see what your LLM tools are sending

#106

Post by jurgenaut23 »

ratg13 wrote: post #97
jurgenaut23 wrote: post #96

Earlier quoted context omitted.

No, you’re in the process of vibe coding stuff you don’t understand and you will most likely never understand until you take the time to open a book.

Your comment contains nothing but insults. This is not a place for you to try and make yourself feel better by disparaging others.

You might find my comment insulting but saying that it contains insults is inaccurate.

Also, OP claims that he is here to learn, but he is mostly chasing cheap GH stars to boost his resume. How insulting is that?

Re: Show HN: A MitM proxy to see what your LLM tools are sending

#107

Post by arowthway »

catlifeonmars wrote: post #103
arowthway wrote: post #102

Earlier quoted context omitted.

Don't use it if you plan to auto accept terminal commands, without a sandbox, while on a public wifi in a cafe, next to a hacker who decides to bet on you running a very niche configuration.

All you need is to manipulate DNS, inject a record with a long TTL and the rest is not required. It scales very well and I guarantee this is not the only instance of misconfigured host verification. In other words, this is not as niche as you might think.

If you're able to manipulate DNS, can't you just issue your own certificate for the domain? Even if it would be revoked moments later, mitmproxy doesnt check it even when ssl_insecure=false:

https://github.com/mitmproxy/mitmproxy/issues/2235

EDIT: Maybe I incorrectly assumed you meant authoritative DNS.

Re: Show HN: A MitM proxy to see what your LLM tools are sending

#109

Post by lionkor »

jmuncor wrote: post #74
catlifeonmars wrote: post #71

Earlier quoted context omitted.

I’m not sure you fully understand the implications of the misconfiguration of mitmproxy there. Effectively you provided an easily accessible front door for remote code execution on a user’s machine. No offense, but I wouldn’t trust anything else you published. I think it’s great that you are learning and it is difficult to put yourself out there and publish code, but what you originally wrote had serious implications…

Ohh my, no offense taken... The next time I will be a lot more careful with the stuff that I put out there. Learning and getting the hang of it, would love if you either comment on the code or here any other things you think could be improved. I am in the process of getting better and appreciate all the blunt and transparent feedback. No one grows out of praise.

I don't think you can get professionals to review code that you didn't even bother typing yourself.

You aren't learning much. You're vibe coding, which means you learn almost nothing, except maybe prompting your LLM better.

Re: Show HN: A MitM proxy to see what your LLM tools are sending

#110

Post by catlifeonmars »

arowthway wrote: post #107
catlifeonmars wrote: post #103

Earlier quoted context omitted.

All you need is to manipulate DNS, inject a record with a long TTL and the rest is not required. It scales very well and I guarantee this is not the only instance of misconfigured host verification. In other words, this is not as niche as you might think.

If you're able to manipulate DNS, can't you just issue your own certificate for the domain? Even if it would be revoked moments later, mitmproxy doesnt check it even when ssl_insecure=false: https://github.com/mitmproxy/mitmproxy/issues/2235 EDIT: Maybe I incorrectly assumed you meant authoritative DNS.

You got it, authoritative not necessary. It just needs to be your router, your ISPs resolver, or the one at your public library/coffee shop/hotel etc. I’d throw BGP route poisoning in there too, but then you have much bigger problems lol.

Like you pointed out in your original post, this would be expensive to run as a targeted attack, but it has good unit economics if you scale it up, wait, and then harvest.

Post reply on HN