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

I've been using Claude Code for a couple of days

twitter.com

Post reply on HN
Print view View original

141–150 of 507 posts

Re: I've been using Claude Code for a couple of days

#141

Post by bglazer »

credit_guy wrote: post #139

I'm using Copilot for writing documentation jupyter notebooks. I do lots of matplotlib plots. Setting up these plots takes lots of repetitive lines of code. Like plt.legend(). With Copilot these lines just show up, and you press tab and move on. Sometimes it is freaky how it guesses what I want to do. For this type of work, Copilot increases my productivity by a factor of 5 easily. There are other types of work where…

Yeah copilot is very good for matplotlib. Clunky interface with lots of repetitive code, but also tons of examples on the internet means that I almost never write matplotlib code by hand anymore.

Re: I've been using Claude Code for a couple of days

#142

Post by istjohn »

mechanicum wrote: post #104
mtlynch wrote: post #33

This is particularly interesting, as Steve Yegge works on (and I think leads) Sourcegraph Cody[0], which is a competitor to Claude Code. Cody does use Claude Sonnet, so they do have some aligned interests, but it's still surprising to see Yegge speak so glowingly about another product that does what his product is supposed to do. [0] https://sourcegraph.com/cody

I mean, doing that is pretty much what made him (semi-)famous in the first place ( https://gist.github.com/chitchcock/1281611 ).

Thanks, I never read that one. Yegge's writing is just delicious. He could write a guide to watching paint dry and I would savor every word.

Re: I've been using Claude Code for a couple of days

#143

Post by intrasight »

chaosprint wrote: post #56

It seems the original poster hasn't extensively tried various AI coding assistants like Cursor or Windsurf. Just a quick heads-up based on my recent experience with agent-based AI: while it's comfortable and efficient 90% of the time, the remaining 10% can lead to extremely painful debugging experiences. In my view, the optimal scenarios for using LLM coding assistants are: - Architectural discussions, effectively re…

> extremely painful debugging experiences.

I'd claim that if you're debugging the code - or even looking at it for that matter - that you're using AI tools the wrong way.

Re: I've been using Claude Code for a couple of days

#144

Post by elcomet »

hleszek wrote: post #55

I must have been a little too ambitious with my first test with Claude Code. I asked it to refactor a medium-sized Python project to remove duplicated code by using a dependency injection mechanism. That refactor is not really straightforward as it involves multiple files and it should be possible to use different files with different dependencies. Anyway, I explain the problem in a few lines and ask for a plan of wh…

I'm wondering if you can prompt it to work like this - make minimal changes, and run the tests at each step to make sure the code is still working

Re: I've been using Claude Code for a couple of days

#145

Post by bob1029 »

I find that maintaining/developing code is not an ideal use case for LLMs and is distracting from the much more interesting ones.

Any LLM application that relies more-or-less on a single well-engineered prompt to get things done is entry level and not all that impressive in the big picture - 99% of the heavy lifting is in the foundation model and next token prediction. Many code assistants are based on something like this out of necessity of needing to support anybody's code. You can't rely on too many clever prompt chaining patterns to build optimizations for Claude Code because everyone takes different approaches to their codebase and has wildly differing expectations for how things should go down. Because the range of expectations is so vast, there is a lot of room to get disappointed.

The LLM applications that are most interesting have the model integrated directly with the product experience and rely on deep domain expertise to build sophisticated chaining of prompts, tool calling and nesting of conversations. In these applications, the user's experience and outcomes are mostly predetermined with the grey areas intended to be what the LLM is dealing with. You can measure things and actually do something about it. What was the probability of calling one tool over the other in a specific context of use? Placing these prompts and statistics alongside domain requirements will enable you to see and make a difference.

Re: I've been using Claude Code for a couple of days

#146

Post by chaosprint »

intrasight wrote: post #143
chaosprint wrote: post #56

It seems the original poster hasn't extensively tried various AI coding assistants like Cursor or Windsurf. Just a quick heads-up based on my recent experience with agent-based AI: while it's comfortable and efficient 90% of the time, the remaining 10% can lead to extremely painful debugging experiences. In my view, the optimal scenarios for using LLM coding assistants are: - Architectural discussions, effectively re…

> extremely painful debugging experiences. I'd claim that if you're debugging the code - or even looking at it for that matter - that you're using AI tools the wrong way.

I'd be very interested to know of a way to make it work with AI that doesn't require debugging if you can illustrate.

Re: I've been using Claude Code for a couple of days

#147

Post by UncleEntity »

hleszek wrote: post #55

I must have been a little too ambitious with my first test with Claude Code. I asked it to refactor a medium-sized Python project to remove duplicated code by using a dependency injection mechanism. That refactor is not really straightforward as it involves multiple files and it should be possible to use different files with different dependencies. Anyway, I explain the problem in a few lines and ask for a plan of wh…

> ...do all the changes myself little by little, checking that the tests keep succeeding as I go along.

Or... you can do that with the robots instead?

I tried that with the last generation of Claude, only adding new functionality when the previously added functionality was complete, and it did a very good job. Well, Claude for writing the code and Deepseek-R1 for debugging.

Then I tried a more involved project with apparently too many moving parts for the stupid robots to keep track of and they failed miserably. Mostly Claude failed since that's where the code was being produced, can't really say if Deepseek would've fared any better because the usage limits didn't let me experiment as much.

Now that I have an idea of their limitations and had them successfully shave a couple yaks I feel pretty confident to get them working on a project which I've been wanting to do for a while.

Re: I've been using Claude Code for a couple of days

#148

Post by deanputney »

trescenzi wrote: post #86

I’m sorry what is happening with this paragraph: > As long as the bank authorizations keep coming through, it will push on bug fixes until they're deployed in production, and then start scanning through the user logs to see how well it's doing. I enjoy using these tools. They help me in my work. But the continual hype makes the discussion around them impossible to be genuine. So I ask, genuinely, did I miss the confi…

I cannot tell if the original tweet is sarcasm or not. Sections like this make me think yes? It's got to be at least tongue-in-cheek.

Re: I've been using Claude Code for a couple of days

#149

Post by MarkMarine »

vander_elst wrote: post #140

Are there any videos showing these very advanced use cases? I'd be interested in learning how to achieve this level of proficiency. At the moment I still feel I'm better off without ai

Claude code doesn’t need magic prompts. It’s not perfect but holy moly is it good, when it’s working it just one shots things for you. It’s just EXPENSIVE. I’ve spent 30$ in tokens on it this week.

Cursor’s “chat with your codebase” is a funny joke compared to Claude Code. Ask it questions, have it figure things out.

I had it analyze the openAPI schema and the backend that is serving the schema for the API I’m writing and write end to end tests for the API. Then I did my normal meetings and it was done with huge chunks of it, it had run the code locally and tested against the actual endpoints to understand if the end to end tests were working or it had found a bug. Then it fixed the bugs it found in the backend codebase. My prompt: “write me end to end tests against my openAPI schema”

That was it. 30$ in tokens later, pressing enter a bunch of times to approve its use of curl, sed, etc…

Re: I've been using Claude Code for a couple of days

#150

Post by roflyear »

abxyz wrote: post #26
develoopest wrote: post #4

I must be the dumbest "prompt engineer" ever, each time I ask an AI to fix or even worse, create something from scratch it rarely returns the right answer and when asked for modification it will struggle even more. All the incredible performance and success stories always come from these Twitter posts, I do find value in asking simple but tedious task like a small refactor or generate commands, but this "AI takes the…

I think it's probably the difference between "code" and "programming". An LLM can produce code and if you're willing to surrender to the LLMs version of whatever it is you ask for, then you can have a great and productive time. If you're opinionated about programming, LLMs fall short. Most people (software engineers, developers, whatever) are not "programmers" they're "coders" which is why they have a positive impres…

> LLMs version of whatever it is you ask for, then you can have a great and productive time

Sure, but man are there bugs.

Post reply on HN