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

Cursor 1.0

cursor.com

Post reply on HN
Print view View original

361–370 of 492 posts

Re: Cursor 1.0

#361

Post by Oreb »

artursapek wrote: post #323
sunaookami wrote: post #169

Earlier quoted context omitted.

>developers are loaded with cash Maybe in the US? I will never pay 100$ for a subscription and I despise that people normalized it by even buying this stuff instead of saying "no, that's way too expensive".

Why not? If you charge $50/hr and it saves you even just two hours a month, it’s a profitable trade.

That “if” doesn’t apply to all of us, though. Not everybody is paid by the hour. I’d love to try something like Claude code, but $100 per month is way too expensive for me, and it probably wouldn’t even give me a single extra dollar of income. I think I’ll just wait for the time when local LLMs will be good enough to be a viable alternative.

Re: Cursor 1.0

#362

Post by mhalle »

therein wrote: post #209
TheYumasi wrote: post #202

Earlier quoted context omitted.

Claude code asks you permissions for every command. It also gives you the possibility of marking commands as safe so next time it can use them without asking .

So these agents that people are so excited about spawning in parallel stop and ask you before executing each command they choose to execute? What kind of life is that. I'd rather do something myself than tell 5 AI agents what I want and then keep approving each command they are going to run. I'm not saying it is better if they run commands without my approval. This whole thing is just doesn't seem as exciting as othe…

I think it's something you have to try in order to understand.

Running commands one by one and getting permission may sound tedious. But for me, it maps closely to what I do as a developer: check out a repository, read its documentation, look at the code, create a branch, make a set of changes, write a test, test, iterate, check in.

Each of those steps is done with LLM superpowers: the right git commands, rapid review of codebase and documentation, language specific code changes, good test methodology, etc.

And if any of those steps go off the rails, you can provide guidance or revert (if you are careful).

It isn't perfect by any means. CC needs guidance. But it is, for me, so much better than auto-complete style systems that try to guess what I am going to code. Frankly, that really annoys me, especially once you've seen a different model of interaction.

Re: Cursor 1.0

#363

Post by Oreb »

input_sh wrote: post #242
anktor wrote: post #172

Does anyone have experience with using this or another agent on local files? No company I know of will approve this for their owned repositories. What about Gitlab instead of GitHub, is there an equivalent to cursor 1.0 product?

If you have a decent GPU or a modern Mac, you can run something like LM Studio ( https://lmstudio.ai/ ) or Ollama ( https://ollama.com/ ), configure the tool such as Cursor to use those models you've downloaded (personally I use Zed.dev), and then everything happens straight on your computer. Responses will be somewhat slower and not as good as state-of-the-art models, but they still can be helpful. Git host doesn't…

What model do you use? I’ve been trying devstral with Zed, and found it rather disappointing.

Re: Cursor 1.0

#364

Post by weego »

rgbrenner wrote: post #337

Recently canceled cursor. I think there's a shift happening right now with the improvements in the ability to process large context sizes and stay on task: Traditional code editing -> autocomplete -> file editing -> agent mode This is basically a gradient of AI output sizes. Initially, with the ability to generate small snippets (autocomplete), and moving up to larger and larger edits across the codebase. Cursor repr…

Yes, my initial distrust lead me to believe that I should only trust whatever ai agent it is to handle the granular issues in code. But they're fundamentally bad at it unless you spend more time prompting and re-prompting than a human with the right context would spend doing it properly.

Now I've changed my technical planning phase to write in a prompt-friendly way, so I can get AI to bootstrap, structure, boilerplate and usually also do the database setup and service layer, so I can jump right into actually writing the granular logic.

It doesn't save me planning or logic overhead, but it does give me far more momentum at the start of a project, which is a massive win.

Re: Cursor 1.0

#365

Post by mhalle »

raincole wrote: post #218
gazagoal wrote: post #154

As someone who has two annual subs of Cursor Pro (one from student account and another from Lenny's newsletter), I just spent $100 on Claude Code and I haven't touched Cursor AI for any coding tasks since. If you already spend anything near or over $100 on Cursor, it's no brainer. The agent experience is night and day. No more wrong tool-calling, premature ending of conversation, failure to apply changes or overwriti…

Claude Code is always quite slow for me. I'm on Windows though so I'm not sure if the performance hit comes from WSL. Anyone experienced differences between Windows/Mac/Linux?

It's slow between each UI interaction compared to auto-complete type systems. But CC can do much more per step. That's especially true if you use it to write scripts that encapsulate multiple steps of a process. Then it can run the scripts.

Perhaps not coincidentally, that's what efficient (or "lazy", you choose) developers do as well.

Re: Cursor 1.0

#366

Post by kaibee »

davedx wrote: post #300
yard2010 wrote: post #248

Earlier quoted context omitted.

That's a really smart observation. It's hard to add sophisticated abstractions though, because they are all selling text by the pounds (kilos?). So it feels the same as vendor lock for a cucumber seller, doesn't it? The seller can sell you an experience that would lock you in, but aside from it there is no moat since anyone can sell cucumbers.

To try and give examples: an autonomous agent that can integrate with github, read issues, then make pull requests against those issues is a step (or maybe two) above an LLM API (cucumber seller). It doesn't take much more of a stretch to imagine teams of agents, coordinated by a "programme manager" agent, with "QA agents" working to defined quality metrics, "architect" agents that take initial requirements and break…

I don't know what the future holds, but I know that this pattern is the 'horseless carriage' of developer automation.

Re: Cursor 1.0

#368

Post by dgfl »

sunaookami wrote: post #169
deadbabe wrote: post #96

Earlier quoted context omitted.

Other professions pay a lot for their tools, and developers are loaded with cash.

>developers are loaded with cash Maybe in the US? I will never pay 100$ for a subscription and I despise that people normalized it by even buying this stuff instead of saying "no, that's way too expensive".

I work in a cleanroom to fabricate semiconductor devices and I spend hundreds of euros per hour to use specific tools which mostly just use electricity and maintenance. Should we complain that it’s too expensive or should we use them because they’re worth the price?

Things have a price for a reason. It’s up to you whether it’s worth paying that or not.

Re: Cursor 1.0

#369

Post by hu3 »

adastra22 wrote: post #151
hu3 wrote: post #23

Earlier quoted context omitted.

Copilot in VSCode has autocomplete and also something they call "next edit". In my experience, next edit is a significant net positive. It fixes my typos and predicts next things I want to do in other lines of the same file. For example, if I fix a variable scope from a loop, it automatically scans for similar mistakes nearby and suggests. Editing multiple array values is also intuitive. It will also learn and sugges…

Cursor does this. And in my experience it gets it perfectly right 95% of the time or better. A lot of times I can start editing something and then just keep hitting tab over and over again until the change is complete--including jumping around the file to make edits in various disconnected places. Of course you can do most of this in Copilot too, but you'd expect something that maybe works and needs a lot of cleanup.…

> Of course you can do most of this in Copilot too, but you'd expect something that maybe works and needs a lot of cleanup.

That hasn't been my experience with Copilot next edit feature.

It often understands exactly what I'm doing and I'm able to just tab tab around the file. Like my example about variable loop scope fixing.

My experience is that Copilot does everything you said including considering files I viewed previously to better understand what I'm trying to do.

Re: Cursor 1.0

#370

Post by nxobject »

vladstudio wrote: post #367

I must mention https://ampcode.com/manual which is my favorite toy of them all right now. Has almost no settings; uses Claude 4, no way to change model; just works! (Unfortunately, not a paid comment.)

A small thing, but I appreciate their free trial usage "starter pack" - the landscape of SWE assistance tools is pretty large these days, and it's impossible to assess fit for the use cases you're interested in without trials. As much as all the positive feedback and/or hype about Anthropic's product is tempting, for hobbyist use I can't quite justify shelling $20 out of the box. (That's to say I can't bill it to someone paying me...)
Post reply on HN