An improved implementation of the Ralph Wiggum technique for autonomous AI agent orchestration

CI-native security testing for MCP servers. Attack simulation, schema drift detection, and health scoring before agents depend on them.