ROMA, an open-source meta-agent framework, has demonstrated superior performance in complex task solving. Built on TechFlow, ROMA uses a hierarchical recursive task tree to efficiently coordinate multi-agent systems. Its search agent, ROMA Search, achieved a 45.6% accuracy on the SEALQA benchmark's Seal-0 subset, surpassing models like Kimi Researcher and Gemini 2.5 Pro. The framework's modular design allows for the integration of various agents, tools, or models, supporting parallel execution. ROMA is particularly suited for applications in finance, research, and content generation.