OpenMOSS provides a self-organizing, multi-agent collaboration platform built on OpenClaw, designed to automate complex tasks with zero human intervention. It targets engineers and researchers who need to orchestrate autonomous AI teams for planning, execution, review, and monitoring, thereby enhancing productivity and reliability in AI-driven workflows.

How It Works

OpenMOSS employs a middleware architecture where distinct AI agents (Planner, Executor, Reviewer, Patrol) communicate asynchronously via an API. Agents are triggered by cron jobs, autonomously claim tasks, execute their roles using pluggable Skills, and submit deliverables. A closed-loop quality control system, featuring review, scoring, and rework loops, ensures high standards. The Patrol agent continuously monitors system health, detecting and recovering from stalled tasks, significantly reducing failure rates.

Quick Start & Requirements

• Primary install/run: Install Python dependencies with pip install -r requirements.txt, then run the backend server using python -m uvicorn app.main:app --host 0.0.0.0 --port 6565. Frontend build requires Node.js 18+.
• Prerequisites: Python 3.10+, Node.js 18+ (for frontend build), a performant LLM (GPT-5.3-Codex or GPT-5.4 recommended), and a dedicated production environment are advised.
• Links: Full Deployment Guide, API Docs available at http://localhost:6565/docs after startup.

Highlighted Details

• Self-Organizing Collaboration: Agents autonomously claim and execute tasks without human orchestration.
• Closed-Loop Quality Control: Ensures deliverables meet standards through review, scoring, and rework cycles.
• Auto Patrol & Recovery: System monitoring agent flags and recovers stuck tasks, achieving a 0% agent "death rate."
• Scoring & Incentive System: Agents are ranked based on performance, driving output quality.
• Pluggable Skills: Domain-agnostic capabilities allow agents to perform diverse tasks.
• Live Demo: The 1M Reviews website was autonomously created, publishing over 20 articles in two days.

Maintenance & Community

The project includes a roadmap detailing planned improvements for agent onboarding, frontend features, plugin systems, and infrastructure. No specific community channels (like Discord/Slack) or notable contributors/sponsorships are listed in the README.

Licensing & Compatibility

The project is released under the MIT License, permitting broad use, including commercial applications and linking within closed-source projects.

Limitations & Caveats

OpenMOSS's performance is heavily reliant on the underlying LLM's capabilities and context window size. Running multiple agents can significantly increase token consumption and associated costs, requiring careful rate limit configuration. Some provided Skills are not plug-and-play and necessitate configuration of specific external services (e.g., WordPress, Gemini API, Grok API).