AI Agent Testing: Systematically validate planning, memory, tool use, and failure recovery

Evaluate an AI agent's planning, memory, tool feedback, and stopping decisions across multi-step tasks and recovery paths.