Key Features

• AI Integration: Uses a KoboldCpp server with the Google Gemma model.
• Flexible Configuration: Can be configured to use OpenAI models through a simple YAML configuration file.
• OS Interaction: Capable of interacting with the operating system, suggesting and executing terminal commands upon user confirmation, see example sessions.
• Optional Wolfram|Alpha Integration: Automatically attempts to augment the LLM’s context via communication with the official Wolfram|Alpha Short Answers API.
• Introspective Contextual Augmentation (ICA): Enhances AI responses by automatically gathering and incorporating relevant contextual information through introspective reasoning. See Introspective Contextual Augmentation for more details.
• Multi-Platform: Runs on Linux, macOS, and Windows, adapting to each environment automatically by collecting context of the running system.
• Dual-Mode Operation: Functions both as a GUI application and a stand-alone terminal tool for quick command generation and embedding of answers in scripts.
• Pastime Mode: Engage in human-like conversations with an AI companion, optionally impersonating specific characters or personalities.
• Intelligent Caching: Implements a sophisticated request caching system for both Wolfram|Alpha and LLM queries, optimizing API usage and response times.