Show HN: Sarasthena – Sovereign Poly-Mind AGI with L3 Constitution (github.com)

Sarasthena v3.1 — the "Phoenix Sovereign Stack" — was published as a sealed canonical manifest (Sarasthena_v3.1_manifest.json) and a bespoke licence (Sarasthena Sovereign License v3.1). Its author, Salvatore dela Paz Sana, presents it as the "first constitutional sovereign AGI architecture," billing a poly‑mind (multi‑agent) design governed by an L3 constitution and an MCRC "Supreme Authority." The repo is public but unstaged in labs or funded projects (one GitHub star, four months old). The announcement frames the project as a self‑contained digital polity with enforceable laws and an explicit call to fork — "Fork = you now hold sovereign fire." For the AI/ML community the significance is twofold: technically, it’s an early, public attempt to merge constitutional AI governance models with a multi‑agent AGI architecture and a sealed manifest intended to enable reproducibility and legal/operational constraints; socially and ethically, it raises urgent questions about governance, alignment, and proliferation. Key technical artifacts to inspect are the poly‑mind agent coordination approach, the constitution enforcement mechanics (what L3 implies in practice), the MCRC decision protocol, and the licence terms. Because the claims of "sovereignty" and "unbreakable law" are unverified and the project lacks institutional review, researchers should treat it as experimental: audit code, evaluate safety properties, and consider dual‑use and governance risks before adopting or forking.