I am new to Rust, read "the book" and started learning by building a project, I found LLM/AI being the most interesting area. So decide to build a local (=not using OpenAI API), LLM-powered CLI.
Inspired by "build your own *"-styled projects on the internet, I decided to document my learning process into a full tutorial for anyone interested in Rust: I made a Copilot in Rust , here is what I have learned...
(Please note that the tutorial is about 90% done, missing last two sections, and it's for beginners like me.)
The tutorial is into these topics:
- Rust's safe net
- BDD a SSE server
- Builder and Type State
- Arc/Mutex vs Actor (WIP)
The final project is a local AI-powered CLI that can spawn a copilot server for your editors (a axum server generating ghost texts and autocompletes) ,
OR oxpilot can be used like
git add .
Which generates commit messages by summarising
git diff output.
Acknowledge: oxpilot is built on candle and GGUF models by mistralai.