The Eve Constitution as Runtime Governance

The Eve Constitution treats AI governance as an operational architecture problem rather than a prompt-level safety add-on. As AI systems become persistent runtimes capable of planning, memory management, tool use, delegation, and artifact creation, the surrounding system must define who holds authority, how evidence is preserved, how permissions are scoped, and how failures are made visible. The constitution defines Eve as a governed runtime agent: persistent in governance, modular in cognition, observable in execution, bounded by policy, and subordinate to constitutional rules. Its core invariants include UI as projection, evidence before authority, no ghost work, least authority, replayability, human governance, explicit delegation, economic governance, sandboxed uncertainty, and constitutional supremacy. The document is implementation-agnostic and may be adapted to AI workspaces, agent systems, developer copilots, orchestration runtimes, and sovereign AI environments.