For some reason setting up agents in a loop with a solid prompt and new context each iteration seems to result in higher quality work for larger or more difficult tasks than the chat interface. It's like the agent doesn't have to spend half its time trying to guess what you want
replyI use Codex CLI or Claude Code
replyI don't even necessarily ask it to fix the bug — just identify the bug
Like if I've made a change that is causing some unit test to fail, it can just run off and figure out where I made an off-by-one error or whatever in my change.
it's more like "this function is crashing with an inconsistent file format error. can you figure out how a file with the wrong format got this far into the pipeline?". in cases like that the fix is usually pretty easy once you have the one code path out of several thousands nailed down.
reply“here's a failing test—do whatever you can to fix it”
replyBad idea. It can modify the code that the test passes but everything else is now broken.
I've heard this said a lot but never had this problem. Claude has been decent at debugging tests since 4.0 in my experience (and much better since 4.5)
replyYou do things like ralph loops.
replyhttps://github.com/snarktank/ralph
Its constantly restarting itself, looking at the current state of things, re-reading what was the request, what it did and failed at in the past (at a higher level), and trying again and again.