AI Agent 'Audit Trail' Dashboard

A middleware tool that sits between LLM agents and the user, providing a visual 'reasoning log' and automated verification checkpoints. It allows users to define success criteria (e.g., 'file saved', 'API returned 200') and pauses the agent if evidence is missing.