Automation Auditor is a governance layer for multi-agent AI systems, built on a deep LangGraph swarm. It observes agent traffic, audits decisions against configurable policies, and enforces guardrails inline — rather than relying on post-hoc log review.
The goal is to make autonomous-agent automation safe to operate at scale: every agent action is policy-checked, every escalation has a reviewable trail, and every unsafe path is short-circuited before it touches production systems.
