Articles
Explanatory analyses of delegated authority, each grounded in tests and commits of the Authority repository.
Articles explain a behavior, a defect or a design choice and link to the evidence behind it. They are not normative: SPEC.md remains the reference for what the engine must do. New articles are published when there is something verified to say, not on a fixed schedule. For the vocabulary, start with What is AI delegation?
Approving One Action Must Not Approve Its Twin
When two agent actions differ only by their identifier, a human approval granted to one must not authorize the other. How Authority binds approvals to action instances, and why the fingerprint-scoped view answers differently.