DongsenZhang/

MSB

This is the official Hugging Face dataset repository for MCP Security Bench (MSB). The MCP Security Bench (MSB) is an end-to-end evaluation suite that systematically measures how well LLM agents resist MCP-specific attacks throughout the full tool-use pipeline: task planning, tool invocation, and response handling. It comprises more than 2,000 attack instances across 12 attack categories.

Paper: https://arxiv.org/abs/2510.15994 Github:… See the full description on the dataset page: https://huggingface.co/datasets/DongsenZhang/MSB.