by someguy101010 on 5/28/25, 9:55 PM with 2 comments
Wanted to take a crack at agents so I wrote one for resolving merge conflicts based on this blog post https://sketch.dev/blog/agent-loop. I have found this to be pretty effective on some prs i've thrown at it but I'd run this in a vm because it just runs shell commands locally.
How does the agent know which to keep and which to discard for a conflict? This would that it has deep contextual information about the codebase it's looking at.