Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A staff engineer's journey with Claude Code

sanity.io

Post reply on HN
Print view View original

111–120 of 410 posts

Re: A staff engineer's journey with Claude Code

#111

Post by utyop22 »

mumbisChungo wrote: post #109
utyop22 wrote: post #83

Earlier quoted context omitted.

I'm finding what's happening right now kinda bizarre. The funny thing is - we need less. Less of everything. But an up-tick in quality. This seems to happen with humans with everything - the gates get opened, enabling a flood of producers to come in. But this causes a mountain of slop to form, and overtime the tastes of folks get eroded away. Engineers don't need to write more lines of code / faster - they need to ge…

>ones thinking power is diminished overtime by interacing with LLMs etc. Sometimes I reflect on how much more efficiently I can learn (and thus create) new things because of these technologies, then get anxiety when I project that to everyone else being similarly more capable. Then I read comments like this and remember that most people don't even want to try.

And? Go create more stuff.

Come back and post here when you have built something that has commercial success.

Show us all how it's done.

Until then go away - more noise doesn't help.

Re: A staff engineer's journey with Claude Code

#112

Post by plaguuuuuu »

swframe2 wrote: post #3

Preventing garbage just requires that you take into account the cognitive limits of the agent. For example ... 1) Don't ask for large / complex change. Ask for a plan but ask it to implement the plan in small steps and ask the model to test each step before starting the next. 2) For really complex steps, ask the model to write code to visualize the problem and solution. 3) If the model fails on a given step, ask it t…

I've been using a few LLMs/agents for a while and I still struggle with getting useful output from it.

In order for it not to do useless stuff I need to expend more energy on prompting than writing stuff myself. I find myself getting paranoid about minutia in the prompt, turns of phrase, unintended associations in case it gives shit-tier code because my prompt looked too much like something off experts-exchange or whatever.

What I really want is something like a front-end framework but for LLM prompting, that takes away a lot of the fucking about with generalised stuff like prompt structure, default to best practices for finding something in code, or designing a new feature, or writing tests..

Re: A staff engineer's journey with Claude Code

#113

Post by yahoozoo »

nostrademons wrote: post #39
swframe2 wrote: post #3

Preventing garbage just requires that you take into account the cognitive limits of the agent. For example ... 1) Don't ask for large / complex change. Ask for a plan but ask it to implement the plan in small steps and ask the model to test each step before starting the next. 2) For really complex steps, ask the model to write code to visualize the problem and solution. 3) If the model fails on a given step, ask it t…

I've found that an effective tactic for larger, more complex tasks is to tell it "Don't write any code now. I'm going to describe each of the steps of the problem in more detail. The rough outline is going to be 1) Read this input 2) Generate these candidates 3) apply heuristics to score candidates 4) prioritize and rank candidates 5) come up with this data structure reflecting the output 6) write the output back to…

How does a token predictor “apply heuristics to score candidates”? Is it running a tool, such as a Python script it writes for scoring candidates? If not, isn’t it just pulling some statistically-likely “score” out of its weights rather than actually calculating one?

Re: A staff engineer's journey with Claude Code

#114

Post by mumbisChungo »

utyop22 wrote: post #111
mumbisChungo wrote: post #109

Earlier quoted context omitted.

>ones thinking power is diminished overtime by interacing with LLMs etc. Sometimes I reflect on how much more efficiently I can learn (and thus create) new things because of these technologies, then get anxiety when I project that to everyone else being similarly more capable. Then I read comments like this and remember that most people don't even want to try.

And? Go create more stuff. Come back and post here when you have built something that has commercial success. Show us all how it's done. Until then go away - more noise doesn't help.

I don't think there's anything I could tell you about the companies I've built that would dissuade you from your perspective that everyone is as intellectually lazy as your projection suggests.

Re: A staff engineer's journey with Claude Code

#115

Post by salty_frog »

swframe2 wrote: post #3

Preventing garbage just requires that you take into account the cognitive limits of the agent. For example ... 1) Don't ask for large / complex change. Ask for a plan but ask it to implement the plan in small steps and ask the model to test each step before starting the next. 2) For really complex steps, ask the model to write code to visualize the problem and solution. 3) If the model fails on a given step, ask it t…

This is my algorithm for wetware llms.

Re: A staff engineer's journey with Claude Code

#116

Post by ryandrake »

abraxas wrote: post #44
lordnacho wrote: post #23

I'm using Claude all the time now. It works, and I'm amazed it worked so easily for me. Here's what it looks like: 1) Summarize what I think my project currently does 2) Summarize what I think it should do 3) Give a couple of hints about how to do it 4) Watch it iterate a write-compile-test loop until it thinks it's ready I haven't added any files or instructions anywhere, I just do that loop above. I know of people…

I tried to follow the same pattern on a backend project written in Python/FastAPI and this has been mostly a heartache. It gets kind of close but then it seems to periodically go off the rails, lose its mind and write utter shit. Like braindead code that has no chance of working. I don't know if this is a question of the language or what but I just have no good luck with its consistency. And I did invest time into de…

What I find helpful in a large project is whenever Claude goes way off the rails, I correct it, and then tell it to update CLAUDE.md with instructions in its own words how to not do it again in the future. It doesn't stop the initial hallucinations and brainfarts, but it seems to be making the tool slowly better as it adds context for itself.

Re: A staff engineer's journey with Claude Code

#117

Post by hn_acc1 »

wordofx wrote: post #20
rvnx wrote: post #12

Earlier quoted context omitted.

Your tips are perfect. Most users will just give a vague tasks like: "write a clone of Steam" or "create a rocket" and then they blame Claude Code. If you want AI to code for you, you have to decompose your problem like a product owner would do. You can get helped by AI as well, but you should have a plan and specifications. Once your plan is ready, you have to decompose the problem into different modules, then make…

> Most users will just give a vague tasks like: "write a clone of Steam" or "create a rocket" and then they blame Claude Code. This seems like half of HN with how much HN hates AI. Those who hate it or say it’s not useful to them seem to be fighting against it and not wanting to learn how to use it. I still haven’t seen good examples of it not working even with obscure languages or proprietary stuff.

When it's best suggestion (for inline typing) is bring back a one-off experiment in a different git worktree from 3 months ago that I only needed that one time.. it does make me wonder.

Now, it's not always useless. It's GREAT at adding debugging output and knowing which variables I just added and thus want to add to the debugging output. And that does save me time.

And it does surprise me sometimes with how well it picks up on my thinking and makes a good suggestion.

But I can honestly only accept maybe 15-20% of the suggestions it makes - the rest are often totally different from what I'm working on / trying to do.

And it's C++. But we have a very custom library to do user-space context switching, and everything is built on that.

Re: A staff engineer's journey with Claude Code

#118

Post by skydhash »

2muchcoffeeman wrote: post #63
lucasyvas wrote: post #55

Earlier quoted context omitted.

I reached this conclusion pretty quickly. With all the hand holding I can write it faster - and it’s not bragging, almost anyone experienced here could do the same. Writing the code is the fast and easy part once you know what you want to do. I use AI as a rubber duck to shorten that cycle, then write it myself.

I’ve been trapped in a hole of “can I get the agent to do this?” And the change would have taken me 1/10th the time. Choosing the battles to pick is part of the skill at the moment. I use AI for a lot of boiler plate, tedious tasks I can’t quite do a vim recording for, small targeted scripts.

How many of these boilerplate do you actually have to do? Any script or complicated command that I had to write was worthy to be recorded in some bash alias or preserved somewhere. But they mostly live in my bash history or right next to the project.

The boilerplate argument is becoming quite old.

Re: A staff engineer's journey with Claude Code

#119

Post by meerab »

block_dagger wrote: post #25

The author doesn't make it clear why they switched from Cursor to Claude. Curious about what they can do with Claude that can't be done with Cursor. I use both a lot and find Cursor to be superior for the very large codebases I work in.

Personal opinion:

Claude code is more user friendly than cursor with its CLI like interface. The file modifications are easy to view and it automatically runs psql, cd, ls , grep command. Output of the commands is shown in more user friendly fashion. Agents and MCPs are easy to organized and used.

Re: A staff engineer's journey with Claude Code

#120

Post by jpollock »

Avoiding the boilerplate is part of the job as a software developer.

Abstracting the boilerplate is how you make things easier for future you.

Giving it to an AI to generate just makes the boilerplate more of a problem when there's a change that needs to be made to _all_ the instances of it. Even worse if the boilerplate isn't consistent between copies in the codebase.

Post reply on HN