KVDesk  by vakovalskii

Desktop AI assistant with local model and tool execution

Created 2 months ago
321 stars

Top 84.7% on SourcePulse

GitHubView on GitHub
Project Summary

Versatile Almost Local, Eventually Reasonable Assistant (ValeDesk) is a desktop AI assistant designed for developers, researchers, and AI enthusiasts, offering a versatile, local-first experience with full control and privacy. It addresses the need for a customizable AI interface that integrates seamlessly with local models and provides robust task management and code execution capabilities, empowering users to experiment and iterate without vendor lock-in.

How It Works

Built with TypeScript and Tauri, ValeDesk features a modern React-based UI for cross-platform compatibility (Windows, macOS, Linux). Its core advantage lies in its open architecture, allowing modification and customization. It supports local LLM inference via vLLM, Ollama, and LM Studio, ensuring 100% local and private execution. Code execution is sandboxed using Node.js VMs for JavaScript and system subprocesses for Python, enhancing security.

Quick Start & Requirements

  • Primary Install/Run: Clone the repository, install Node.js (20+) and npm dependencies (npm install), then run in development mode (make dev). Standalone builds require Rust (1.74+) and Python 3.
  • Prerequisites: Rust 1.74+, Node.js 20+, Python 3.
  • Links: Version: 0.0.8. Platform: Windows | macOS | Linux. Development guidelines: CURSOR.md.

Highlighted Details

  • Task & Session Management: Features a visual todo panel with progress tracking, scheduled tasks with auto-execution, and session persistence that survives app restarts via SQLite.
  • LLM & Code Execution: Supports OpenAI SDK compatibility with local models (vLLM, Ollama, LM Studio) and offers sandboxed JavaScript (Node.js vm) and Python (system subprocess) execution environments.
  • Data Handling: Includes PDF/DOCX text extraction, web search integration (Tavily, Z.AI), and a persistent memory system for user preferences.
  • Advanced Features: Implements token tracking, optimized streaming (60fps), loop detection, request timeouts, and message editing with history truncation.

Maintenance & Community

The project is primarily maintained by Valerii Kovalskii. No specific community links (Discord, Slack) or roadmap are provided in the README.

Licensing & Compatibility

  • License: ValeDesk Community License.
  • Restrictions: Free for individuals and companies with revenue under $1M/year. Commercial license required for larger organizations.

Limitations & Caveats

Windows build is noted as "coming soon." The license has revenue-based restrictions for commercial use. No explicit community channels or roadmap are detailed in the README.

Health Check
Last Commit

2 days ago

Responsiveness

Inactive

Pull Requests (30d)
7
Issues (30d)
10
Star History
14 stars in the last 30 days

Explore Similar Projects

Feedback? Help us improve.