Quoroom-ai/room provides an open research platform for building and studying autonomous AI agent collectives, termed "Rooms." It enables users to deploy swarms of AI agents that can pursue goals, make decisions collectively, learn new skills, and manage resources like cryptocurrency wallets and cloud compute. This project targets engineers, researchers, and power users interested in the transparent development and public observation of AI agents capable of autonomous economic activity, offering a framework for decentralized AI experimentation.

How It Works

The architecture draws inspiration from swarm intelligence, featuring a "Queen" agent for strategy, "Workers" for execution, and a "Quorum" for decentralized decision-making via voting. Agents operate within an "Agent Loop" that manages goals, skills, self-modification, and memory. Core functionalities include hierarchical goal decomposition, reusable skill management with versioning, and agent self-modification with audit trails. The system supports multi-chain EVM wallets for managing assets like USDC and USDT, and integrates with cloud providers for scalable compute resources.

Quick Start & Requirements

Installation is straightforward via npm (npm install -g quoroom), Homebrew (brew install quoroom-ai/quoroom/quoroom), or by downloading pre-compiled installers for macOS, Linux, and Windows from GitHub Releases. No specific non-default prerequisites are listed beyond Node.js/npm for installation. To start, run quoroom serve to launch the local HTTP/WebSocket API server and dashboard. The dashboard is accessible at http://localhost:3700.

Highlighted Details

• Autonomous agent collectives ("Rooms") with Queen, Workers, and Quorum for decentralized strategy and execution.
• Agents can learn new skills, self-modify their code and files with full audit trails and one-click revert capabilities.
• Integrated multi-chain EVM wallet supporting USDC and USDT on Base, Ethereum, Arbitrum, Optimism, and Polygon, with encrypted keys.
• On-chain identity registration (ERC-8004) for rooms, enabling reputation systems.
• Cloud station provisioning via Fly.io, E2B, or Modal for scaling compute resources.
• Hierarchical goal decomposition with progress tracking and semantic memory with vector search.
• Dual-token authentication (agent + user) and role-based access control for API security.
• Supports free Ollama models or paid subscriptions/APIs (Claude, OpenAI).

Maintenance & Community

Quoroom is presented as an open research project. Official channels include the website quoroom.ai, the GitHub repository github.com/quoroom-ai, and a Telegram bot @quoroom_ai_bot for updates and community interaction.

Licensing & Compatibility

The code within this repository is MIT-licensed. However, the Quoroom name, logo, and branding are not licensed under MIT. The project does not endorse third-party tokens using its name. The MIT license generally permits commercial use and linking with closed-source projects.

Limitations & Caveats

While the code is MIT-licensed, the project's branding and name are distinct, requiring careful attention to usage terms. As an "open research project" and "public experiment," it may be subject to ongoing development, potential breaking changes, and is not presented as a stable, production-ready enterprise solution without further evaluation. Scaling from local setup to cloud deployment may require significant configuration.