06 · AGENT ACCESS / GOVERNANCE
Agent Authority
A verification layer for what AI agents can actually access and do.
THE SYSTEM
A focused project around testing the real authority boundary of AI agents. The central idea is to verify observed access and actions rather than assuming that declared permissions or intended behavior equal actual capability.
ARCHITECTURE
Declared capability → observed access → action verification → evidence → authority boundary.
EVIDENCE STATE
Verified GitHub repository