arterm-cli

A terminal AI coding agent that runs local models — Ollama, any OpenAI-compatible endpoint (LM Studio, vLLM, …), or a .gguf loaded in-process. Rich Ink TUI, live streaming, an interactive model picker, and permission-gated file/shell tools sandboxed to your working directory.

Install

Requires Node.js >= 22.

npm install -g arterm-cli
# or: npx arterm-cli

Quick start

ollama serve            # start a model backend (https://ollama.com)
ollama pull qwen2.5:7b
arterm --model qwen2.5:7b

Inside the TUI: type to chat, Enter to send, ? for help, Alt+P to pick a model, Esc to cancel a turn, Ctrl+C to quit. write/edit/bash ask for permission before running (use --yolo to skip).

arterm                 # interactive chat (default)
arterm models          # list models across providers
arterm pull <model>    # download a model via Ollama

Full documentation, providers, security model, and config: https://github.com/Arclude/Arterm-CLI

License

MIT © Arclude