BonsaiCLI

Terminal workspace for chatting with Bonsai AI models locally.

Description

BonsaiCLI is a terminal-based interface for interacting with Bonsai AI models. It provides a clean, keyboard-driven experience with four core surfaces: Chat for conversations, Setup for installation and health checks, Settings for configuration, and Commands for CLI entrypoints.

Features

• Interactive TUI — Keyboard-driven terminal interface with streaming responses
• Multiple Surfaces — Chat, Setup, Settings, and Commands views
• Runtime Management — Automatic detection and installation of AI runtimes
• Model Catalog — Browse and install available Bonsai models
• Slash Commands — Quick actions via / prefix commands
• Persistent History — SQLite-backed conversation storage

Prerequisites

• Go 1.23 or higher
• Linux or macOS (Windows support in progress)
• For GPU acceleration: NVIDIA CUDA, AMD ROCm, or Apple Metal

Installation

Build from Source

./build.sh

Install

./install.sh ./BonsaiCLI

The install script places BonsaiCLI in ~/.local/bin and attempts to update your shell PATH.

Note: The scripts provide actionable error messages for common issues: missing Go, missing binary, non-writable install path, or missing shell profile permissions.

Quick Start

# Open the chat interface
BonsaiCLI chat

# Open setup to install runtime and models
BonsaiCLI setup

# Check system health
BonsaiCLI doctor

Running BonsaiCLI without arguments opens the chat interface.

Commands

Command	Description
BonsaiCLI chat	Open the chat interface
BonsaiCLI setup	Runtime installation and health
BonsaiCLI settings	Configure defaults
BonsaiCLI doctor	Print system diagnostics
BonsaiCLI models	List available models
BonsaiCLI help	Show help information

First-Time Setup

Run the setup command to configure your environment:

BonsaiCLI setup

The setup view displays:

• Detected platform and architecture
• Runtime backend (cuda, rocm, cpu, metal) and support level
• Support reason and suggested remediation
• Runtime installation state
• Python diagnostics and huggingface_hub availability
• Installed and missing Bonsai models

Use Ctrl+S or Enter inside the setup view to install the runtime, bootstrap the app-managed Python downloader when needed, and install the selected model.

Keyboard Shortcuts

Shortcut	Action
Tab	Switch between Chat, Setup, Settings, Commands
Ctrl+N	Create a new conversation
Ctrl+P	Cycle the active conversation model
Ctrl+S	Install or start the runtime
Ctrl+,	Open settings
F1 or ?	Open command reference
Esc	Return to chat
Ctrl+C or Ctrl+D	Quit

Slash Commands

Command	Description
/new	Start a new conversation
/model	Show current model
/model <name>	Switch to a specific model
/rename <name>	Rename current conversation
/setup	Open setup view
/settings	Open settings view
/goto chat	Return to chat
/help	Show help

Settings

Configure default behavior in the settings view:

• Default Model — Model used for new conversations
• Auto-Start — Whether the runtime starts automatically before requests

Settings are stored in the local SQLite-backed app state.

Environment Support

The runtime automatically detects your OS, architecture, and available tooling to select an appropriate backend profile.

Supported Configurations

Platform	Backend	Status
Linux x86_64 + NVIDIA CUDA	CUDA	✅ Supported
Linux x86_64 (no NVIDIA)	CPU	⚠️ Degraded
Linux AMD/ROCm	ROCm	⚠️ Degraded
macOS	Metal	⚠️ Experimental

Unsupported combinations are reported in setup and doctor with suggested fixes.

Diagnostics

Doctor Command

BonsaiCLI doctor

Prints active directories, runtime health, Python availability, and installed models.

Models Command

BonsaiCLI models

Displays the bundled model catalog with install state, size, and VRAM requirements.

License

This project is provided as-is for educational and personal use. See the source code for details.