This repository demonstrates a variety of **MCP Poisoning Attacks** affecting real-world AI agent workflows.