Hacker News Reader

Comment by MuffinFlavored | Hacker News Reader

I have a question. I do not like the concept of "agent mode" for AI. I'm a control freak and I want to control every line that gets committed because I am responsible for it and need to understand/visualize/memorize every part of codebases I work on.

Is Claude Code good for the "ask" flow? No, right?

The old flow before agent mode got added. Select some code, ask questions about it or give an instruction on editing it and then choose to accept the change.

As I understand (I could be wrong), with agent mode, it edits the file for you, no way for you to accept before it does, so you have to manually check the diff, roll back parts you don't want, etc.

Am I right?

View on HN

It could be better. I think the PMs and investors and decision makers at these companies are running with a "we want to replace / automate developers" philosophy, while these tools are actually best at augmenting developers. And so they're sorta builing with this "I'll do everything and ask you for confirmation" (and basically encourage you to give me blanket permission).

In reality these tools would be best if they took a more socratic method, a more interactive pair programming approach. So instead of giving you a blanket diff to accept or refuse or "No, and here's some changes" -- it should be more dialog oriented.

Of all of them so far though, I think Claude Code is closest to this. IF you prompt it right you can have a much more interactive workflow, and I find that most productive.

Claude Code is definitely more agentic, but you can use it in a variety of ways. In Plan Mode it won't touch the code, and by default it asks you to accept every single diff. With IDE integration you can definitely just highlight some code and ask questions, I don't see why that workflow wouldn't work.

I like aider's solution of encapsulating each edit in a git commit; I hope that gets widely adopted.

Add two lines to CLAUDE.md and claude code can do this as well :)

Unfortunately aider is showing its age. It is great for what it does, but better LLMs + "agentic" have shown that you can get more in the swe domain.

There was a paper recently where they had an LLM evolve tool harnesses and got ~20% more than w/ aider on the benchmark they used, so it's pretty clear that the models + tools (+better harness) are better than just aider.

1) You can plug in any model into aider 2) It can be quite agentic

> evolve tool harnesses

Claude code & Gemini cli etc. don't do this either

Don't get me wrong, I love aider and I've been using it since the early days. I'm just commenting on the overall "gains" and imo they are higher with the latest tools (claude code, gemini, etc).

As for 1), I agree but you force the model to work within aider's constraints. Claude4 for example excels at the agentic flow and it's better at that than providing the diffs that aider expects.

As for the last sentence, I disagree. They are evolving the stack, and more importantly they are evolving both at the same time, stack + LLM. That's the main reason they all subsidise use atm, they are gathering data to improve both. If I were to place a bet right now, I'd say that provider_tool + provider_LLM > 3rd party tool + same model in the short, medium and long term.

Oh, that's a good point, I misunderstood you to mean: The LLM writes it's own harnesses etc.

Each edit as in every little change?

Yes. Then you squash to make real commits.

We tend to allow this. But you can review the diff before you allow it. It seems easier to say "roll that back" or hit esc if you see it doing things you dont like and then correcting it. I have to say the velocity is insane of coding this way. We also commit a LOT and explicitly dont allow claude to push code ever. That way we can roll back if needed, but honestly it's pretty rare to need to. The MAX plan is a must to work this way though.

> As I understand (I could be wrong), with agent mode, it edits the file for you, no way for you to accept before it does, so you have to manually check the diff, roll back parts you don't want, etc.

> Am I right?

With cursor you get reasonably flexible control at many levels. You can have it only suggest changes that you have to apply manually or you can have it make automatic changes with various ways to review, change, reject or accept. I usually have the changes made automatically but don’t accept the changes automatically. Cursor has a UI that lets you review each edit individually, for the whole file or all files. Depending on the situation I will use whichever level is appropriate. The UI also allows you to revert changes or you can ask the AI to undo or rework a change that you just approved so there’s plenty of ways to do large changes without giving up control. There’s also a stop button you can use to interrupt mid-stream if the work it’s doing isn’t what you want. It isn’t flawless but I haven’t found myself in a corner where I couldn’t get back to a happy path.

> As I understand (I could be wrong), with agent mode, it edits the file for you, no way for you to accept before it does, so you have to manually check the diff, roll back parts you don't want, etc.

You’re sort of technically correct but I wouldn’t really describe it this way exactly. You have to proactively accept or reject all changes to your files in some way.

It is almost impossible to accidentally commit code you don’t want.

It’s not really an edit in the same sense as an unstated change. It doesn’t even really do that until you accept the result.

It’s basically saving you a UI step compared to ask mode with basically no downside.

Reply on news.ycombinator.com