Show HN: Magenta.nvim – AI coding plugin for Neovim focused on tool use (opens in new tab)

(github.com)

64 pointsanonymid1y ago8 comments

I've been developing this on and off for a few weeks. There are a few videos on the README page showing demos of the plugin.

I just shipped an update today, which adds: - inline editing with forced tool use - better pinned context management - prompt caching for anthropic - port to node (from bun)

Check it out!

8 comments

vessenes1y ago

Watched the video. I like this! Really smart idea to get us one step closer to workable human workflows.

Watching you work with it, I found I wanted something that mushes aider and your stuff together. Where I am spending time in aider, I am spending time complaining about claude using bad patterns, assuming bad structure/types/etc or misunderstanding the purpose of a set of code. Where it looks like you are spending time is in fixing coding errors, and manually reviewing patches.

At the very least, it seems like automating / making easy compilation and type checking results would be nice.

Having used aider for over a year now, I know he's spent a lottt of time on prompting customization for both improved code quality and also DIFF format, and I wonder if you would benefit from some of those lessons in terms of getting better code out of different models.

Anyway, this is awesome, and I love the idea of giving some tools to the LLM to engage with the codebase, pull context and then code. Super cool.

anonymidOP1y ago

There are lsp tools available to the agent, so in theory it should be able to ask for types, references and diagnostics.

In using it I've found that the agent doesn't really make use of these tools unless I instruct it to. I think I need to do some messing around with prompts to encourage the agent to use them more.

I think the next major thing I want to add is commands to allow you to send symbol types / diagnostics to the agent before it asks for them, which should help speed up some of these workflows.

Thanks for the pointer to take another look at aider - the DIFF format sounds really interesting. Though it seems like the license is Apache 2.0 so I think I'll have to learn a bit about what that means since I made mine MIT licensed.

para_parolu1y ago

Are the any plans to add local models? Ot tools only work with openai and claude?

whimsicalism1y ago

How does this compare to avante.nvim?

and does it support r1 endpoints?

dimtion1y ago

Both those questions are answered clearly in the readme:

> compared to avante > I think it's fairly similar. However, magenta.nvim is written in typescript and uses the sdks to implement streaming, which I think makes it more stable. I think the main advantage is the architecture is very clean so it should be easy to extend the functionality. Between typescript, sdks and the architecture, I think my velocity is pretty high. I haven't used avante in a while so I'm not sure how close I got feature-wise, but it should be fairly close, and only after a couple of weeks of development time.

And:

> Another thing that's probably glaringly missing is model selection and customization of keymappings, etc...

whimsicalism1y ago

thank you - missed that section at the bottom on my phone

bashtoni1y ago

r1 doesn't support function calling, so I'd assume the answer will be no.

anonymidOP1y ago

The interface for a provider does abstract over this, so you could implement it in the way that most other plugins do by rolling your own prompts and parsing: https://github.com/dlants/magenta.nvim/blob/main/node/provid...

But yeah... it probably would take substantially more effort and be more brittle than having the model providers do it for you.

I'm happy to accept PRs if folks want to go for it :)

j / k navigate · click thread line to collapse

8 comments

vessenes1y ago

Watched the video. I like this! Really smart idea to get us one step closer to workable human workflows.

At the very least, it seems like automating / making easy compilation and type checking results would be nice.

Anyway, this is awesome, and I love the idea of giving some tools to the LLM to engage with the codebase, pull context and then code. Super cool.

anonymidOP1y ago

There are lsp tools available to the agent, so in theory it should be able to ask for types, references and diagnostics.

In using it I've found that the agent doesn't really make use of these tools unless I instruct it to. I think I need to do some messing around with prompts to encourage the agent to use them more.

I think the next major thing I want to add is commands to allow you to send symbol types / diagnostics to the agent before it asks for them, which should help speed up some of these workflows.

para_parolu1y ago

Are the any plans to add local models? Ot tools only work with openai and claude?

whimsicalism1y ago

How does this compare to avante.nvim?

and does it support r1 endpoints?

dimtion1y ago

Both those questions are answered clearly in the readme:

And:

> Another thing that's probably glaringly missing is model selection and customization of keymappings, etc...

whimsicalism1y ago

thank you - missed that section at the bottom on my phone

bashtoni1y ago

r1 doesn't support function calling, so I'd assume the answer will be no.

anonymidOP1y ago

But yeah... it probably would take substantially more effort and be more brittle than having the model providers do it for you.

I'm happy to accept PRs if folks want to go for it :)

j / k navigate · click thread line to collapse