01Integrate the Model Context Protocol (MCP) for standardized context provision to LLMs and seamless component reuse.
02Rapidly reproduce and extend RAG modules via a standardized MCP architecture, encapsulating components as reusable Servers and function-level Tools.
03Construct complex RAG pipelines with low code using YAML for control structures (sequential, loop, conditional branching).
04747 GitHub stars
05Unify evaluation and comparison with built-in workflows and metric management, supporting 17 mainstream scientific benchmarks.
06Natively support multi-structured pipeline workflow control, defining logic at the YAML level for decoupled execution.