kuse-ai/kuse_cowork

What it solves

Kuse Cowork is a local-first, model-agnostic desktop agent designed to handle complex document and file management tasks while maintaining strict privacy and security. It provides an open-source alternative to Claude Cowork, allowing users to run AI agents on their own machine with full control over which LLM provider they use.

How it works

Built with Rust and Tauri, the application runs natively on macOS, Windows, and Linux. Users provide their own API keys (BYOK) for providers like Anthropic, OpenAI, or connect to local models via Ollama or LM Studio. The agent operates within a designated workspace folder and uses Docker containers to isolate command execution, ensuring that AI-generated commands do not compromise the host system.

Who it’s for

It is designed for users who want a private, secure AI assistant for automating document processing, research synthesis, and data analysis without relying on a single proprietary ecosystem.

Highlights

• Model Agnostic: Supports Claude, GPT, and local models via OpenAI-compatible endpoints.
• Secure Execution: Uses Docker for containerized isolation of external commands.
• Crosspatform Native: Written in Rust for high performance and low memory footprint (~10MB app size).
• Extensible Skills: Includes a system for custom capabilities, with default support for docx, pdf, pptx, and xlsx files.
• MCP Support: Integrates the Model Context Protocol for seamless tool expansion.