slop post about slop code
He closed his laptop. The work kept going.
Six months of writing code exclusively with agents
81–90 of 110 posts
Re: Six months of writing code exclusively with agents
#82Re: Six months of writing code exclusively with agents
#83> AI agents are just software, after all. I wasn’t going to understand them by reading prompt guides. I had to use them for real work, see where they failed, change the prompts, tools, or environment, and try again. AI agents are just software, so the only way to understand them is trial and error?
Tried and true approach. Sometimes the only way to proceed if you are dealing with - embedded devices, protocols, long and deep stacktraces or freecad.
Re: Six months of writing code exclusively with agents
#84I feel a really odd dissonance with all these accounts. I have access to top models via Cursor at work. Every time I think to myself, here’s a tedious but well defined task, with clear success criteria, where you can achieve a lot with persistent iterations. I’ll give it to Claude to do. First, it takes me ages to describe the task properly. All these clear success criteria, well, instead of writing code from a clear…
Take a step back and give it the essential functional requirements. Or, consider what information is driving your extremely detailed instructions and give it the raw source if you can. Meeting transcripts, messy notes on some adhoc conversation, etc. The more you try to micromanage it, the more impossible to work with it it is. Of course not every situation calls for just giving it high level requirements. But not ev…
Well, quite. Recently I gave it a database dump, and told it what the processed data looks like. Since the raw data was messy, there was a lot to explain.
It didn’t go well.
Re: Six months of writing code exclusively with agents
#85Several comments here touch on the core problem: agents are writing more code than we can review. Seniors have never had enough time to review, and the prolific output from coding agents is making it worse. Moreover, the code is almost always good. So you're reviewing a tsunami of pretty good code, which means you get review fatigue and the whole thing just becomes theater.
For me, this means the checks have to be more than just "I looked at it". There are two things that happen before I ever see it: first, as others have mentioned, I have a model from a different family review it with well-defined criteria. Same-family reviewers share bias, so it must be a different one. Second, I have a core set of deterministic gates (like lint, but also unit tests) that must run. In either case, failures go back to the coding agent.
And only then do I bother. But I really don't read everything. If it is bog-standard CRUD operations, the agents are pretty good at that, especially if they are using mature packages. I focus on critical things, like how it enforces permissions.
This works well for me, though it leaves one major issue untouched: whether this is worth building. The agents tend to be a bit overeager, so I have to do a lot of work up front to make sure the output will add value. The gates can only check the artifact, not my intent.
Re: Six months of writing code exclusively with agents
#86Earlier quoted context omitted.
I've heard this reported, but it's hard to imagine it. The biggest bottleneck for me has always been hammock time[1]. 1. https://github.com/matthiasn/talk-transcripts/blob/master/Hi...
Yes, but once you have the solution crystalized in your head do you not feel some frustration in not getting it implemented as soon as possible? I often found myself impatient in getting something I knew would work to actually work, and it was always due to the tiresome fractal of complexities that any non-trivial task involves, like looking up an API or working around the quirks of some dependency. On a related note…
I simply type it out equanimously. The attachment to speed is neither here nor there. It will end up on the screen in due time.
Re: Six months of writing code exclusively with agents
#87Re: Six months of writing code exclusively with agents
#88Earlier quoted context omitted.
Yes, but once you have the solution crystalized in your head do you not feel some frustration in not getting it implemented as soon as possible? I often found myself impatient in getting something I knew would work to actually work, and it was always due to the tiresome fractal of complexities that any non-trivial task involves, like looking up an API or working around the quirks of some dependency. On a related note…
I simply type it out equanimously. The attachment to speed is neither here nor there. It will end up on the screen in due time.
I think you're talking about the flow state, which makes sense. I've said it before there was an aspect to programming that was almost meditative, that zone where ideas flow easily into code. I do miss that experience. (I also suspect the lack of it contributes to the burnout reported by heavy AI coders.)
Unfortunately, most modern software development is atrocious, full of distractions and incidental complexity and tangential issues that are not conducive to staying in the flow. Latent bugs and dependency quirks and environment issues that force you into some irrelevant sidequest were the most frustrating. Like, this video is the prefect analogy for modern software development: https://www.youtube.com/watch?v=AbSehcT19u0
I do NOT miss that experience and I'm glad to let the agents take care of it.
Re: Six months of writing code exclusively with agents
#89Earlier quoted context omitted.
Yes, but once you have the solution crystalized in your head do you not feel some frustration in not getting it implemented as soon as possible? I often found myself impatient in getting something I knew would work to actually work, and it was always due to the tiresome fractal of complexities that any non-trivial task involves, like looking up an API or working around the quirks of some dependency. On a related note…
I simply type it out equanimously. The attachment to speed is neither here nor there. It will end up on the screen in due time.
Sure, if I had enough time.
But I have more software I want to write than I have time, between my day job, my kids and family, chores, other hobbies, etc.
I had so many projects I wanted to do, but when I thought about all of the work that went into each one, I realized I didn't have time to do them and they were pushed to the back burner.
These last few months I have been able to get to so many of those back burner projects because I don't need nearly as much time anymore.
Re: Six months of writing code exclusively with agents
#90I also haven't written more than a couple lines of code in months, but I definitely do read the code. You have to, or at least provide lots of Claude Rules to help steer it. AI has too many bad coding practices by default. But you don't need to modify lines of code by hand, you can just tell it what to fix. I'm predicting a Docker for AI Coding. There really is a ton of stuff you need to set up to do AI coding/vibe c…
"You can just tell it what to fix"
For a lot of small changes, I am waaaaaay faster than an LLM.