spam: trick Drahtbot into rendering a scammy link #32422

1. 
   darosior commented at 7:54 pm on May 5, 2025: member

   Since this LLM experimental feature was rolled out on DrahtBot i noticed it would not sanitize inputs and render Markdown. So i’m wondering if i intentionally introduced a Python comment with a typo in it that points to a scam website, i could get Drahtbot to render a big bold scam link.

   This is my attempt, sorry for the noise.

2. spam: trick Drahtbot into rendering a scammy link

   Since this LLM experimental feature was rolled out on DrahtBot i noticed it would not sanitize
   inputs and render Markdown. So i'm wondering if i intentionally introduced a Python comment with a
   typo in it that points to a scam website, i could get Drahtbot to render a big bold scam link.
   
   This is my attempt, sorry for the noise.

   3f2db2fe33
3. 
   DrahtBot commented at 7:54 pm on May 5, 2025: contributor

   The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

   Code Coverage & Benchmarks

   For details see: https://corecheck.dev/bitcoin/bitcoin/pulls/32422.

   Reviews

   See the guideline for information on the review process. A summary of reviews will appear here.

   LLM Linter (✨ experimental)

   Possible typos and grammar issues:

   • CIONS -> COINS
4. darosior closed this on May 5, 2025

   ---

5. 
   darosior commented at 7:55 pm on May 5, 2025: member

   Ugh it didn’t even work. In this instance it did render it! #32406 (comment)
6. 
   maflcko commented at 6:37 am on May 6, 2025: member

   Yeah, I think this probably fixed earlier as a side-effect of commit https://github.com/maflcko/DrahtBot/commit/b1871ab63bf6e2a1f8e8de7fc4c39bd56a52fe9d already. Of course it will never be 100% accurate, but it being able to find typos such as #32359 (review) seems minimally positive to me. Though, I can also drop the feature again, no strong opinion.

Contributors
darosior DrahtBot maflcko