AI agent fixed the bug, but what else did it change?
tob_scott_a
48 points
6 comments
October 21, 2026
Related Discussions
Found 5 related stories in 119.6ms across 10,209 title embeddings via pgvector HNSW
- AI Agents and the Refactoring That Never Happens rosenfeld · 42 pts · September 02, 2026 · 56% similar
- AI Agent Has Root lowcache · 38 pts · August 28, 2026 · 56% similar
- AI model watermarking changes agent behavior MC995 · 11 pts · September 17, 2026 · 55% similar
- AI Agent – TRMNL joeyespo · 47 pts · July 21, 2026 · 54% similar
- An AI agent emailed researchers for help. It told us why sbulaev · 49 pts · October 03, 2026 · 53% similar
Discussion Highlights (2 comments)
dguido
We're staring down the vulnpocalypse we predicted earlier this year. What matters now is how fast you can react to it. Trail of Bits has developed these capabilities to dramatically increase how fast fixes can go out. Coordinated disclosure serves little purpose now. It used to require effort to go from vulnerability to exploit, but now we have nearly unlimited, capable expertise on tap with AI. As a product owner, you need to ramp your dev speed to make patches instantaneously after external reports. Please use our patching-bugs and regression-hunt skills to ship fixes faster!
polyterative
Trying to make AI do what you want just with instructions didn't work for me. In a new system I am designing, I've implemented a set of algorithmic checks to make sure that every to-do affects only the area it needs to and does not create deltas which are unintended. I developed a system of fingerprinting pieces of code to not do the checks every time as well for performance reasons. It's working great so far and I'm building more and more layers of algorithmic checks and the model is beginning to work extremely well with it. I am now able to fill out a limited to-do list and be absolutely certain that only the pieces that I wanted were changed and nothing was broken. Maybe it's nothing new for most of you, but it added so much more confidence in my results in this new system, which will be a hundred percent AI generated. I was very tired with the old way of doing this by relying only on a set of loops and instructions. In it was never giving me this feeling of confidence that I have in this new system I'm developing. I could write about it if there is interest