4 ms·

I love the accept reject flow because I still constantly have to stop AI models from writing awful architecture or reimplementing code we already wrote elsewher

I love the accept reject flow because I still constantly have to stop AI models from writing awful architecture or reimplementing code we already wrote elsewhere

flyingoat 4mo ago
Yeah, I have found the same. A lot of times it does get things right, but if it deviates man it can just drift hard. For example, sometimes Claude just obsessively reads files and goes on massive tangents. Then when I stop it and ask, "why are you doing that?", it kindly apologizes and admits it shouldn't have gone on a tangent. The token burn if I don't stop it would be quite high. Granted, this might be because I'm not giving it optimal prompt/negative-prompt instructions though.
chamomeal 4mo ago
I just check the git diff after claude code writes stuff. Stage things before letting it run wild so I can undo whatevs.
- noworriesnate 4mo ago
  That's expensive though. The sooner you stop it from acting out the less you spend on a rabbit trail.
tclancy 4mo ago
How is it different from Keep / Discard in other tools? I've been slowly converting my git repositories to jj locally because that gives me more granular fallback and mix and match options.
- digitaltrees 4mo ago
  Well I tried CLaude Code for the first time in a while (I am building my own coding app www.propelcode.app so I can code on my phone when I take my kids to classes and such) and it literally ignored my question and suggestion and just kept coding away.
imtringued 4mo ago
I hate the accept reject flow, because I want a conventional code review workflow where I can write comments on specific lines of code and maybe edit the code myself. If I reject, then the AI will struggle to modify just the parts I disagree with, if I accept, the AI will tend towards adding code rather than updating the bad code. At that point copy paste without agentic coding tends to work much better.

I love the accept reject flow because I still constantly have to stop AI models from writing awful architecture or reimplementing code we already wrote elsewhere

flyingoat 4mo ago

Yeah, I have found the same. A lot of times it does get things right, but if it deviates man it can just drift hard. For example, sometimes Claude just obsessively reads files and goes on massive tangents. Then when I stop it and ask, "why are you doing that?", it kindly apologizes and admits it shouldn't have gone on a tangent. The token burn if I don't stop it would be quite high. Granted, this might be because I'm not giving it optimal prompt/negative-prompt instructions though.

chamomeal 4mo ago

I just check the git diff after claude code writes stuff. Stage things before letting it run wild so I can undo whatevs.

noworriesnate 4mo ago
That's expensive though. The sooner you stop it from acting out the less you spend on a rabbit trail.

tclancy 4mo ago

How is it different from Keep / Discard in other tools? I've been slowly converting my git repositories to jj locally because that gives me more granular fallback and mix and match options.

digitaltrees 4mo ago
Well I tried CLaude Code for the first time in a while (I am building my own coding app www.propelcode.app so I can code on my phone when I take my kids to classes and such) and it literally ignored my question and suggestion and just kept coding away.

imtringued 4mo ago

I hate the accept reject flow, because I want a conventional code review workflow where I can write comments on specific lines of code and maybe edit the code myself. If I reject, then the AI will struggle to modify just the parts I disagree with, if I accept, the AI will tend towards adding code rather than updating the bad code. At that point copy paste without agentic coding tends to work much better.