Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

VSCode with extensions Copilot [autocomplete] + CLINE [AI chat] + FOAM [obsidian-esk markdown support] is goat. There's no way a closed-source alternative to going to compete with this.


Why not? What does it do that Copilot Agent or Junie can't? All the competitors have a similar UX and the same selection of models.


Have you given the alternatives a genuine test? My experience with Aider (never used Cline) is that it's nowhere near as good


The benefit of Aider is that you can configure a very involved static analysis toolchain to edits which directly triggers new edits in response, and everything is a git commit so it's easy to revert bad edits quickly. I have used both and I find Aider provides more control and produces code faster due to leaner prompts (it's also easier to run multiple Aider instances than Cursor instances), while Cursor has a prettier interface, and I do like being able to see diffs live in files (though I almost never spend the time reading them to accept/reject). I imagine if you don't spend any time configuring Aider cursor would probably seem far better.


what are the most useful changes you've made to the configuration? This could be it - I haven't played with that a whole lot


Create a file like conventions.md in the root of your repository with specific commands for common tasks: running tests, linters, formatters, adding packages Set this as part of the files it reads on startup. Then ask aider to look at your codebase and add to it :) Aider has a lot of slash commands to familiarize yourself with. Ask and web are crucial commands to get the most out of it.

My recommendation to anyone is to use ask the most then tell it to “implement what we discussed” when it looks good.

Hope that helps


The biggest thing is to set it to autofix lint/test issues, then to set up a really good lint/test config. Also, I find that Aider's default system prompt setup is a little less preconfigured out of the box than Cursor's, so it helps to have detailed styleguide/ai rules documents that are automatically added to the chat. I usually configure my projects to add README.md, STYLEGUIDE.md (how to structure/format code) and AIRULES.md (workflow stuff, for instance being socratic with the user when requirements aren't clear or the prompt contains ambiguity, general software engineering principles/priorities, etc).


Are you saying Aider isn't as good as Cursor, or that Cursor isn't as good as Aider?


aider isn't as good


In my limited tests, Aider and CLINE are very similar, but it's really hit/miss depending on the specific task.


Doesn't this rely completely on the AI it is using and not the client?


What does "is goat" mean?


Goat is slang for “(the) greatest of all time.”


I was today years old minus maybe a few months when I learned this, and I had seen it referenced so many times.


This one is not particularly new. I wanna say that GOAT was new in the early 90s.


But not widespread until recently. I'm a 90's kid and I only remember it being used for sports superstars and such.


I was not commenting on the vintage of this, I was commenting on my ignorance which is where the parent or grandparent comment started from. Sorry if it did not come out right


You are one of today's lucky 10,000. https://xkcd.com/1053/


wasn't ali the goat in 70s?



what model do you use with cline?


copilot autocomplete? my experience with it has been very delusional, cursor prediction in cursor(bad naming let's be honest) is simply unmatched


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 suggest formatting prefences and other things such as API changes.

Sure, sometimes it suggests things I don't want but on average it is productive to me.


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. The cursor autocomplete is, more often than not, EXACTLY what you would have hand crafted, without any deficiencies.

It's also somehow tracking what I look at, because I can look up an API and then return to the original source file and the first autocomplete suggestion is exactly what I was looking up, even though there would be no context to suggest it.

It's really quite magical, and a whole different level from Copilot.


> 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.


Reading up on this, it sounds like Copilot adopted the methodology that Cursor has been using internally for more than a year. Which is great, but if your question is "why is everyone using Cursor?" it is because many initially used Cursor when they were the only ones with this feature. I, for example, specifically switched from Copilot on VSCode to Cursor because of the spooky accuracy of Cursor's tab complete compared to Copilot, at the time. This was only a few months ago.


Cursor also does this.


Cursor tab is remarkable. There's a lot of competition for agents but I don't think any other product comes close to their tab completion. Admittedly it might be rather useless in the near future with how things are going though.


Did you change CompletionModel to 'gpt-4o-copilot'? - it may be the default now, provided you keep copilot extension updated.


FOAM / obsidian is markdown + graph building, or does it also add additional keywords to markdown?


I used Cline and Claude Code extensively for a project. Claude code is much better.


Interesting. For me Cline and Roo are king. I would use them exclusively if I could afford it. With Copilot Pro+ it goes a long way but still ends in rate limits down the road


Cline copying features of Claude Code seems like sustainable competition.


Do you use foam for your notes or for llm memory?


foam is just for notes and MARP slides


i just can't keep up with all the new words


FOAM?






Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: