This project provides a desktop application for local, uncensored AI model interaction, including chat, agent capabilities, and image/video generation. It targets users prioritizing privacy, data control, and flexibility, offering a unified interface that supports numerous local and cloud-based AI backends without requiring Docker or API keys. The primary benefit is enabling powerful AI functionalities entirely on the user's hardware, free from external censorship or data collection.

How It Works

The application is built using Tauri v2, featuring a Rust backend and a React/TypeScript frontend, compiled into a single, standalone executable for Windows. It employs a "Plug & Play" setup that auto-detects up to 12 local AI backends (e.g., Ollama, LM Studio, llama.cpp) and provides one-click installation links for any missing ones. For image and video generation, it integrates with ComfyUI, allowing users to download and manage models directly within the app. Key differentiators include its "uncensored by default" stance and advanced agentic features.

Quick Start & Requirements

• Primary Install: Download the .exe or .msi installer from the GitHub Releases page.
• Build from Source: Clone the repository, run npm install, then npm run dev for development or npm run tauri build for a production binary.
• Prerequisites: Windows 10/11 is officially tested and supported. Linux and macOS builds are possible from source but not officially supported. GPU is recommended for image/video generation.
• Links: Website, Discord, GitHub Releases.
• Note: Antivirus software may flag the installer as suspicious due to heuristics on unsigned NSIS installers; verification steps are provided in the SECURITY.md file.

Highlighted Details

• Unified Interface: Supports 12+ local backends (Ollama, LM Studio, vLLM, etc.) and multiple cloud providers (OpenAI, Anthropic, Groq, etc.) with per-conversation switching.
• Advanced AI Features: Includes a Coding Agent (Codex) with file tree access, 14 Agent Tools (web search, shell, code execution), Thinking Mode for AI reasoning visualization, and File Upload + Vision capabilities.
• Multimedia Generation: Integrated image and video generation via ComfyUI, supporting models like FLUX, Z-Image, Wan, and AnimateDiff, with full parameter control and no content filters.
• Privacy & Control: Operates entirely locally with no cloud dependency, data collection, or mandatory API keys.

Maintenance & Community

The project is actively maintained, with recent commits and releases. Community support and discussion are available via a dedicated Discord server. Contribution guidelines and an open roadmap are provided on GitHub.

Licensing & Compatibility

The project is licensed under the AGPL v3. This is a strong copyleft license, meaning any derivative works or modifications distributed must also be made available under the AGPL v3 license. This may impose restrictions on integration with proprietary or closed-source commercial products.

Limitations & Caveats

Official support and testing are limited to Windows 10/11; building and running on Linux or macOS requires compiling from source and may encounter platform-specific issues. Some antivirus programs may generate false positive warnings for the installer.