Deterministic Execution Authority for Autonomous Space Systems: Governance Architecture for Orbital Platforms Under the Outer Space Treaty

Starlink performed roughly 300,000 autonomous collision avoidance maneuvers in 2025, every one decided by onboard AI and none reviewed by a human, while the Outer Space Treaty assigns liability to the launching state with no way to prove the system stayed inside its authorized envelope. This paper specifies an evidence-bound authority kernel for orbital platforms, governing eight operation classes from collision avoidance to deorbit, with receipt chains that scale to 42,000-satellite constellations. Liability defense becomes cryptographic proof instead of organizational assertion.