lean

What this is

This is a Lean 4 proof workbench for small, auditable kernels about admissibility.

The common question is not “can this system act?” but:

when does what has been presented actually license what follows?

Most authorization systems ask whether a caller has the right role, token, or permission. These proofs are about a different failure surface: cases where evidence, standing, freshness, authority, or transition get mistaken for more than they can support.

The payload is not the authority. A system that cannot tell an assertion of authority from the real thing will eventually launder one into the other: stale credentials read as fresh, a dead surface read as healthy, a guess promoted to a verdict, a proxy mistaken for the thing it replaced.

Most modules follow the same pattern:

  1. name the boundary;
  2. model the thing being offered as support;
  3. model the thing it is being used to justify;
  4. prove that the unsupported move cannot cross the boundary.

Some modules audit claims from the Δt research series. Others form the admissibility kernels themselves: authority, standing, freshness, surface authorization, witness invariance, state transition, execution, and corrective layers — a set of small kernels, not a unified calculus.

The point is not to prove an entire institution, platform, incident, or distributed system correct. The point is smaller and more useful: make category errors mechanically visible before they become architecture.

This is a proof workbench, not an oracle.