llm-lsp

Language Server Protocol for Large Language Models

Editors will be able to use this as lsp for code completions.

Current Features:

[x] Code completion using codeium.ai
[x] Uses async
[x] Allows users to use same binary for different AI models with cli option
[x] Saves multiple configs to be used by different instances of llm-lsp

How to build:

cargo build --release

Usage:

Display help

llm-lsp -h

Display version

llm-lsp -V

Generate Config

Once llm-lsp is executed, it will guide through the process to generate API_KEY (in case of codeium) and save the relevant configurations in OS specific toml configs.

llm-lsp generate-config

Configure editor command

llm-lsp server -p codeium

Helix Editor configuration

[language-server.llm-lsp]
command = "llm-lsp"
args = ["server", "-p", "codeium"]

[[language]]
name = "rust"
language-servers = [
    "rust-analyzer",
    "llm-lsp",
]

Future:

[ ] More Documentation
[ ] Add chat support on cli