🦙 LazyLlama

LazyLlama is a lightweight, fast Terminal User Interface (TUI) client for Ollama. It is designed for running local AI models with minimal overhead and intuitive, Emacs-inspired controls directly in your terminal.

✨ Features

• Real-time Streaming: Responses are generated live, providing immediate feedback.
• Markdown Support: Automatic formatting for headers, lists, and bold text.
• Code Highlighting: Syntax blocks are visually separated with custom borders and background colors.
• Smart Scrolling: * AUTOSCROLL: Automatically follows the AI output.
  • MANUAL SCROLL: Locks the view (🔒) when you use PageUp/Down, allowing you to read previous messages undisturbed.
• Model Management: Easily switch between installed Ollama models using arrow keys with separate input/output buffers per model.
• Smart Buffer Management: Each LLM maintains its own chat history, input text, and scroll position.
• Automatic Logging: Every chat session is automatically saved as a text file in ~/.local/share/lazyllama/ (both combined and per-model histories).
• Performance: Built with Rust and Ratatui for ultra-low latency and minimal resource footprint.

🚀 Installation

Prerequisites

• Rust (Stable)
• Ollama (must be running in the background)

Build from Source

1. Clone the repository:

   git clone https://github.com/Pommersche92/lazyllama.git
   cd lazyllama

2. Install it system-wide:

cargo install --path .

⌨️ Controls

Optional debug:

• LAZYLLAMA_DEBUG_KEYS=1 shows key/scroll/render info in the status bar.

🛠 Project Structure

The project follows a modular design for easy maintainability:

• main.rs: Entry point and terminal event handling.
• app.rs: State management and Ollama API integration.
• ui.rs: Rendering logic and Markdown parsing.
• utils.rs: File system operations and session logging.

📖 Documentation

You can generate the full technical documentation locally:

cargo doc --no-deps --open

📄 License

This project is licensed under the GPL-2.0-or-later. See the LICENSE file for details.

📝 Changelog

v0.3.0 - February 2026

• 🧹 Clean Chat Redraw: Clears the chat area before rendering to prevent leftover characters when scrolling
• ⌨️ Input Cursor: Horizontal cursor navigation with Left/Right and a blinking caret in the input field
• 🏁 Home/End Navigation: Jump to start/end of the input line
• 🧭 Word-wise Movement: Ctrl+Left/Right moves by words with smart separators
• 🗑 Word Deletion: Ctrl+Backspace deletes previous word, Ctrl+Delete deletes next word
• ⌦ Delete Key Support: Delete removes the character after the cursor
• 🧪 Dev Key Debug Mode: Optional status bar debug via LAZYLLAMA_DEBUG_KEYS=1

v0.2.0 - February 2026

• 🎯 Per-Model Buffer Management: Each LLM now maintains separate input buffers, chat histories, and scroll positions
• 🔄 Smart Model Switching: Arrow keys now seamlessly switch between models while preserving individual states
• 💾 Enhanced Logging: Separate history files are saved for each model on application exit
• 🎨 Improved UI: Model list shows buffer status indicators and current model highlighting
• 🪟 Windows Compatibility: Fixed double character input issue on Windows by filtering key event types
• 📖 Translated Documentation: Documentation in source code AI-translated from german to english
  • Please report any gramatical errors, AI weirdness and/or other inaccuracies in the github issues.

v0.1.0 - Initial Release

• Basic TUI interface for Ollama
• Real-time streaming responses
• Markdown and code highlighting
• Smart scrolling with autoscroll/manual modes
• Model selection and automatic logging

Developed with ❤️ in the black forest.