[AI Summary]: ROMA is a meta-agent framework that uses recursive hierarchical structures to build high-performance multi-agent systems. The framework processes tasks through a recursive plan-execute loop, breaking down complex problems into parallelizable components while maintaining transparency for easy debugging and iteration. Built on AgnoAgents with Python/FastAPI backend and React/TypeScript frontend, it supports any LLM provider via LiteLLM and includes features like E2B sandboxes for secure code execution and enterprise S3 mounting.
- Developer: Sentient AGI
- License: MIT
- Platform: Python 3.12+, Docker, Web