opencowork  by Safphere

AI desktop assistant for local operations

Created 1 month ago
263 stars

Top 96.9% on SourcePulse

GitHubView on GitHub
Project Summary

OpenCowork is an open-source desktop AI assistant designed to transform a PC into a powerful, locally-controlled work companion. It supports a wide array of agent-capable AI models without vendor lock-in, offering direct access to local file systems and terminal execution. This project targets developers and power users seeking a customizable, privacy-focused AI assistant for learning and advanced task automation.

How It Works

The core architecture is model-agnostic, integrating with any AI supporting agent capabilities (e.g., MiniMax, Claude, GPT) via ClaudeCode compatible interfaces, rather than standard OpenAI SDKs. It features a robust Skills System and MCP Protocol, enabling built-in functionalities like web search, image understanding, and coding plan generation. OpenCowork facilitates deep interaction by allowing read, write, create, and modify operations on local files and the execution of command-line operations, all managed within multi-session contexts for concurrent task handling.

Quick Start & Requirements

  • Installation: Download and install platform-specific packages for Windows and Linux. macOS users may need to bypass security warnings for unsigned applications (e.g., right-click "Open").
  • Build: Source compilation requires bun install && npm run build.
  • Prerequisites: None explicitly listed beyond build tools.
  • Documentation: Configuration Guide, Development Guide available.
  • Source Code: github.com/Safphere/opencowork

Highlighted Details

  • Model Agnostic: Seamless integration with various agent models.
  • Extensible Skills System: Includes 11 built-in Skills and 10 MCP Services for advanced capabilities.
  • Local File & Terminal Control: Enables direct manipulation of local files and command-line execution.
  • Floating Ball UI: Customizable hotkey (Alt+Space) for quick access.

Maintenance & Community

The project acknowledges partnerships with MiniMax (CN & Intl), Zhipu AI, and ZAI (Intl), and actively welcomes collaborations with AI model providers. No direct links to community forums (e.g., Discord, Slack) are provided in the README.

Licensing & Compatibility

Licensed under the Apache License 2.0, generally permitting commercial use. However, the project's disclaimer states it is provided "as-is" for learning and development, with developers disclaiming liability for any losses, which may require careful consideration for production environments.

Limitations & Caveats

A significant risk notice highlights that the AI operates on local file systems and terminals, potentially leading to accidental file deletion or incorrect command execution, alongside prompt injection vulnerabilities. macOS users must navigate security prompts for unsigned applications. The software is explicitly provided for learning and development, with developers disclaiming liability for any resulting damages.

Health Check
Last Commit

3 weeks ago

Responsiveness

Inactive

Pull Requests (30d)
0
Issues (30d)
4
Star History
29 stars in the last 30 days

Explore Similar Projects

Feedback? Help us improve.