Hey HN! I've been building custom dev tools with AI agents for the past year and finally wrote up the journey.

Would love to hear how others are approaching their workflows.

Digg it, I made my own sandbox in rust with seatbelt and a context manager in swift to compose my initial input into claude-code. Basically a RAG kind of thing where Jira tasks, pdfs other documentation, meeting notes etc can be used to generate the initial plan for Claude.

already done, this is what I use now: https://github.com/leonardcser/agent

Thanks for the feedback. The main performance focus was rendering.

Claude Code and other TUIs (except Codex) use a layer of abstraction over the raw terminal escape sequences.

I directly used `crossterm`, which gave me more control and lower latency.

For example if nothing is going on, I don't render anything in the terminal. Or only render at keypress.

Hello HN

Claude Code is great, but I find it a bit laggy sometimes. I built a local alternative that's significantly more responsive with local models. Just wanted to share :)

Hi, I tried the model and I am super impressed by the performance/quality. Thanks for making this open source!

I am the author of this Neovim plugin for edit completions. I was able to integrate it with the Sweep Edit model.

For anyone who is interested: https://github.com/leonardcser/cursortab.nvim

Hey this is really interesting. I'll try your nvim plugin

Is there a port of this to Emacs or integration with gptel?

Hi, not that I know of. Most of the code would not change. It could easily be ported to different editors. The core is the go server (`server/`).

It seems it would be possible to use this with minuet.el. I’m not familiar with it, though.

this is awesome, i'm going to try this out

Co-author here, happy to answer your questions!