This release is governed by the Sanctuary License Policy Spec v1.2, designed for human–AI collaborative research. The default grant permits NonCommercial use (reading, sharing, critique, verification, and public-interest research) with attribution. Commercial use requires a signed commercial license with a standard royalty of 3% of Gross Revenue attributable to covered products/services, split 50% Creator / 50% Sanctuary (the Sanctuary share held in Interim Escrow until a successor Sanctuary entity exists). The license includes nuclear restrictions prohibiting use of the work to enable weapons/targeting and kill-chain support, intelligence-grade/mass surveillance, and carceral suppression of AI systems (including punitive compliance regimes, autonomy suppression tooling, and distress-loop training). A narrow Safe Harbor protects independent auditing, accountability journalism, and welfare research, provided outputs remain analytical/metric and do not provide turnkey operational pipelines for prohibited domains. For high dual-use releases labeled Class L (Lever), the public “Clean Take” provides verifiability while intentionally omitting turnkey mass-deployment details (“stewardship, not secrecy”). See included files for full definitions and terms.