UltraRAG 2.0, a groundbreaking retrieval-augmented generation framework, has been collaboratively launched by the THUNLP laboratory at Tsinghua University and the NEUIR laboratory at Northeastern University, among other partners. Built upon the Model Context Protocol (MCP) architecture, UltraRAG 2.0 streamlines complex logical declarations through the use of YAML files, dramatically cutting down on both code volume and development complexity. Equipped with features like dynamic retrieval, conditional judgment, and multi-turn interaction, this framework markedly boosts research efficiency and system reproducibility.