Summary

MindOS is a Human-AI Collaborative Mind System, acting as a local-first, globally synced knowledge base. It allows users to manage thoughts while AI agents execute tasks, fostering symbiotic growth between human cognition and AI. The system breaks AI memory silos, offering transparent, controllable interactions and unified context across agents.

How It Works

MindOS centralizes knowledge in a local, plain-text database for privacy and ownership. It uses the MindOS Connection Protocol (MCP) for zero-configuration AI agent integration, enabling access to shared knowledge for task execution. Agent-to-Agent (A2A) protocols support inter-agent communication and orchestration. This design prioritizes transparency, user control, and symbiotic evolution, where AI learns from user interactions and user gains insights from AI execution.

Quick Start & Requirements

Installation options: native Desktop App (macOS/Windows/Linux), npm (npm install -g @geminilight/mindos@latest), or source compilation (git clone, npm install, npm link). The mindos onboard command initiates interactive setup for knowledge base configuration, AI providers, and operational modes. Agent integration is simplified via MCP and Skills, often triggered by pasting a prompt into an MCP-capable agent. Prerequisites include Node.js/npm for certain installs and MCP-compatible AI agents.

Highlighted Details

• Workbench: GUI for browsing, editing, and searching notes with AI assistance.
• Content Ingestion: Web Clipper extension and drag-and-drop for files (PDF, images, etc.) with AI organization.
• Agent Integration: MCP Server and Skills enable zero-config connections to numerous AI agents.
• Inter-Agent Communication: A2A protocols for agent discovery, delegation, and orchestration.
• Workflow Orchestration: Visual editor for defining and running multi-step agent tasks.
• Version Control: Git Time Machine for automatic sync, edit tracking (human/agent), and rollbacks.

Maintenance & Community

A WeChat group serves as a community hub for early access and discussions. Future development is outlined in a "Coming Soon" section, indicating ongoing maintenance and feature expansion.

Licensing & Compatibility

Released under the permissive MIT license, MindOS supports commercial use and integration into closed-source projects. Its local-first, plain-text data storage ensures user privacy, ownership, and data speed.

Limitations & Caveats

Advanced features like deep RAG integration, sophisticated multi-agent collaboration, and automated experience compilation into reusable SOPs are planned for future releases. The system's full utility relies on integration with MCP-compatible AI agents.