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

What makes Claude Code so damn good

minusx.ai

Post reply on HN
Print view View original

311–320 of 322 posts

Re: What makes Claude Code so damn good

#311

Post by hamandcheese »

gervwyk wrote: post #76
hamandcheese wrote: post #66

Earlier quoted context omitted.

> easier to validate This is essential to productivity for humans and LLMs alike. The more reliable your edit/test loop, the better your results will be. It doesn't matter if it's compiling code, validating yaml, or anything else. To your broader question. People have been trying to crack the low-code nut for ages. I don't think it's solvable. Either you make something overly restrictive, or you are inventing a very…

Good point. i’m making the assumption that if the LLM has a more limited feature space to produce as output, then the output is more predictable, and thus faster to comprehend changes. Similar to when devs use popular libraries, there is a well known abstraction, therefore less “new” code to comprehend as i see familiar functions, making the code predictable to me.

I think we are essentially describing the same thing. You just want to achieve it by constraining the input space at a significantly higher level (yaml schema defines the output space instead of a compiler and/or test suite).

I still think you'll be at a significant disadvantage since the LLM has been trained on millions of lines of all mainstream languages, and 0 lines of gervwyks funny yaml lang.

Re: What makes Claude Code so damn good

#312

Post by dgunay »

monkpit wrote: post #174
bilekas wrote: post #170

Earlier quoted context omitted.

I would argue this approach will help you in the long term with code maintainability. Which I feel will be one of the biggest issues down the line with AI generated codebases as they get larger.

The solution is to codify these sorts of things in prompts and tool use and gateways like linters etc. you have to let go…

I have been doing this, and it does sort of work, but the problem is that for things that can't easily be turned into deterministic lints, prompting isn't 100% reliable. Every bit you go against the LLM's training data, it's more likely to forget to do it.

Re: What makes Claude Code so damn good

#313

Post by rapind »

tomashubelbauer wrote: post #274
dtagames wrote: post #252

Earlier quoted context omitted.

The difference between Claude Code and Cursor is that one is a command line tool and the other an IDE. You can use Claude models in both and all these techniques can be applied with Cursor and its rules, too. It's Coke vs. Pepsi.

There is also Cursor Agent CLI which is a TUI exactly like CC. I switched to it because I don't like GUI AI assistants, but I also couldn't stand CC always being overloaded and having many bugs that were affecting me. I'm not on Cursor Agent CLI with GPT5 and happy to have an alternative to CC.

How is Cursor Agent + GPT5 holding up? I've been on Claude code for a while, but there's been an increase in timeouts and slowdowns recently.

Re: What makes Claude Code so damn good

#314

Post by tomashubelbauer »

rapind wrote: post #313
tomashubelbauer wrote: post #274

Earlier quoted context omitted.

There is also Cursor Agent CLI which is a TUI exactly like CC. I switched to it because I don't like GUI AI assistants, but I also couldn't stand CC always being overloaded and having many bugs that were affecting me. I'm not on Cursor Agent CLI with GPT5 and happy to have an alternative to CC.

How is Cursor Agent + GPT5 holding up? I've been on Claude code for a while, but there's been an increase in timeouts and slowdowns recently.

For my personal projects it has completely replaced the need for CC with Anthropic models for me. At work, I am waiting for native Windows support. I don't like using AI assistants via WSL. Since both CA and CC are Node apps and CC has since shipped native Windows support, I don't foresee it taking CA long either. Especially since it can be hacked to work that way today as I've experimented with here: github.com/TomasHubelbauer/cursor-agent-windows

Re: What makes Claude Code so damn good

#315

Post by rapind »

tomashubelbauer wrote: post #314
rapind wrote: post #313

Earlier quoted context omitted.

How is Cursor Agent + GPT5 holding up? I've been on Claude code for a while, but there's been an increase in timeouts and slowdowns recently.

For my personal projects it has completely replaced the need for CC with Anthropic models for me. At work, I am waiting for native Windows support. I don't like using AI assistants via WSL. Since both CA and CC are Node apps and CC has since shipped native Windows support, I don't foresee it taking CA long either. Especially since it can be hacked to work that way today as I've experimented with here: github.com/Toma…

Good to know. I’ll try it out on an upcoming greenfield. Thanks.

Re: What makes Claude Code so damn good

#316

Post by dang »

foobarbecue wrote: post #93
1zael wrote: post #82

I've literally built the entire MVP of my startup on Claude Code and now have paying customers. I've got an existential worry that I'm going to have a SEV incident that will trigger a house of falling cards, but until then I'm constantly leveraging Claude for fixing security vulnerabilities, implementing test-driven-development, and planning out the software architecture in accordance with my long-term product roadma…

[flagged]

This is a belated reply, but you broke the site guidelines badly here. It's not ok to attack other users like this, no matter how right you are or feel you are. We ban accounts that do, so please don't do it again.

It's fine, of course, to make your substantive points thoughtfully, but that is a very different kind of comment.

https://news.ycombinator.com/newsguidelines.html

Re: What makes Claude Code so damn good

#317

Post by foobarbecue »

dang wrote: post #316
foobarbecue wrote: post #93

Earlier quoted context omitted.

[flagged]

This is a belated reply, but you broke the site guidelines badly here. It's not ok to attack other users like this, no matter how right you are or feel you are. We ban accounts that do, so please don't do it again. It's fine, of course, to make your substantive points thoughtfully, but that is a very different kind of comment. https://news.ycombinator.com/newsguidelines.html

Apologies, forgot to check my reaction.

Re: What makes Claude Code so damn good

#318

Post by fragmede »

majormajor wrote: post #181
fragmede wrote: post #160

Earlier quoted context omitted.

LLMs write python and typescript well, because of all the examples in their training data. But what if we made a new programming language whos goal was to be optimal for an LLM to generate it? Would it be closer to assembly? If we project that the future is vibe coded, and we scarcely look at the outputted code, testing, instead, that the output matches the input correctly, not looking at the code, what would that la…

What is it that you think would make a certain non-Python language "more optimal" for an LLM? Is there something inherently LLM-friendly about certain language patterns or is "huge sets of training examples" and "a robust standard library" (the latter to conserve tokens/attention vs having to spit out super-verbose 20x longer assembly all day) all "optimality" means?

It's fair to point out that I didn't define exactly what we're optimizing for, but I can have the LLM generate assembly if I ask. It'll be faster than python, at the expense of readability, but if we're no longer writing code, then why not straight up use assembly? ARM vs x86 becomes an issue, but are there other reasons not to use assembly?

Re: What makes Claude Code so damn good

#319

Post by fragmede »

metrix wrote: post #166
fragmede wrote: post #160

Earlier quoted context omitted.

LLMs write python and typescript well, because of all the examples in their training data. But what if we made a new programming language whos goal was to be optimal for an LLM to generate it? Would it be closer to assembly? If we project that the future is vibe coded, and we scarcely look at the outputted code, testing, instead, that the output matches the input correctly, not looking at the code, what would that la…

I have thought the same thing. How is it created? is it an idea by an LLM to make the language, or a dev to create a language designed for an llm. How do we get the LLM to gain knowledge on this new language that we have no example usage of?

Same way we do for any other language. Give it docs and a runtime, and have it go off and generate code and use that training data.

Re: What makes Claude Code so damn good

#320

Post by seivan »

rapind wrote: post #313
tomashubelbauer wrote: post #274

Earlier quoted context omitted.

There is also Cursor Agent CLI which is a TUI exactly like CC. I switched to it because I don't like GUI AI assistants, but I also couldn't stand CC always being overloaded and having many bugs that were affecting me. I'm not on Cursor Agent CLI with GPT5 and happy to have an alternative to CC.

How is Cursor Agent + GPT5 holding up? I've been on Claude code for a while, but there's been an increase in timeouts and slowdowns recently.

[dead]
Post reply on HN