A preserved reference project exploring whether AI agents need an independently governed capability namespace — stable capability identity, cryptographic trust, and cross-boundary coordination not controlled by any single platform. Preserved as a completed study artifact; see the Project Retrospective.