Agent Governance · Method
Each engagement reconstructs the same six questions against the specific workflow under review. The method is consistent; what it finds is not assumed in advance.
Who is formally allowed to do what.
What effects become reachable through composition.
What information became a premise for the decision.
What changed the agent over time.
What happens when authority is removed or execution fails.
Who is accountable and whether they can actually repair the downstream consequence.
Where the method stops
Finding generation, confidence calibration, internal benchmark and falsifier library, and the cross-engagement pattern library used to build each map stay internal to aiclavis. What a client receives is the reconstruction for their own workflow — not the method's internal machinery.
The outcome may be a closed control, a proven gap, or an unresolved boundary. We do not promise which.