🚀 Lume — Local AI Desktop Client

Lume is a high-performance, privacy-first AI desktop chat application for Linux. It provides a clean, modern interface for your local Ollama models — no cloud, no telemetry, no compromise.

Built with Tauri v2, Svelte 5, and Rust, Lume runs blisteringly fast on low-end hardware while delivering a premium user experience that rivals ChatGPT and Claude's desktop apps.

✨ Features

Core

• Zero-Latency Conversations — Connects directly to Ollama on localhost:11434 for fully offline inference
• Real-Time Streaming — Watch tokens appear word-by-word as the CPU/GPU generates them
• Multi-Session Chat — Create, switch between, search, and delete independent chat sessions
• Persistent History — SQLite backend with foreign-key cascading ensures conversations survive restarts
• Model Selection — Switch between any installed Ollama model on the fly

Interface

• Collapsible Sidebar — Premium sidebar with session list, search bar, and smooth 300ms collapse animation
• Smart Titles — Session titles auto-generate from your first message (truncated to 35 chars)
• Time-Relative Stamps — Sidebar shows "Just now", "2h ago", "Yesterday" etc.
• Dark / Light Mode — Native system-preference detection with manual toggle
• Markdown & Syntax Highlighting — Full GFM rendering with highlight.js code blocks
• Scroll-to-Bottom FAB — Floating action button with unread badge and bounce animation

  Interaction & Details

• Thinking Process Parsing — Natively extracts <think> reasoning metadata and renders inline as a collapsible component (for DeepSeek/Gemma reasoning models).
• Edit & Regenerate — Modify previous prompts to branch out chats, or easily regenerate an AI output with one click.
• Stop Generation — Mid-stream capability to instantly abort API outputs, saving CPU tokens.
• Copy Message — One-click text copying to your system clipboard.
• Model Analytics — Subtly displays Generation Time (seconds) and Token Count underneath model responses.

Settings & Personalization

• Settings Panel — Dedicated modal with 5 distinct configuration tabs (Models, Chat, Appearance, Data, About)
• Granular Control — Modify AI temperatures and assign specific models to unique chats
• Agent Personalities — Built-in selectable personas (Coder, Writer, Pirate, Analyst, Assistant)
• Data Portability — Easily Export and Import complete markdown/JSON chat histories natively
• Side Panel Upgrades — Pin favorite chats to the top, toggle bulk deletion modes, and view user profiles dynamically

Performance

• ~5 MB binary size
• < 0.5s startup time
• < 20 MB idle RAM usage
• No Electron — Native WebKit via Tauri

📸 Screenshots

UI Overhaul Complete - Fresh Interface Screenshots Updating Soon

🛠️ Tech Stack

💻 System Requirements

📦 Installation

Pre-built RPM (Fedora)

sudo dnf install ./lume-0.5.0-1.x86_64.rpm

From Source

Prerequisites: Node.js 18+, Rust 1.70+, and Tauri system dependencies:

sudo dnf install webkit2gtk4.1-devel openssl-devel curl wget file \
  libappindicator-gtk3-devel librsvg2-devel pango-devel

Build steps:

git clone https://github.com/sandeep4513m/Lume.git
cd Lume
npm install
npm run tauri dev      # Development mode with hot-reload
npm run tauri build    # Production RPM bundle

Note: Ensure Ollama is installed and running before launching Lume:

systemctl start ollama

🗃️ Project Structure

lume/
├── src/                    # Svelte frontend
│   ├── routes/+page.svelte # Main chat + sidebar UI
│   ├── components/         # Markdown renderer
│   └── lib/ollama.js       # Ollama API client
├── src-tauri/              # Rust backend
│   ├── src/db.rs           # SQLite schema + Tauri commands
│   └── src/lib.rs          # App bootstrap + command registration
├── package.json
└── README.md

🤝 Contributing

Contributions are welcome! Feel free to open issues or submit pull requests.

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.

Built with 💚 using Tauri + Svelte + Ollama