What AI believes
Memory, RAG, policy, workflow truth
Surfaces 2–4 — safe-to-trust before action basis.
Security layer on TraceScript
TraceScript programs substrate state. TrustScript gates what may become trusted enterprise state. Domain surfaces — SmartPeers, CSD/DENSITY, and application whitepapers — run on the same receipt-backed runtime.
TrustScript is organized around what AI trusts, says, and does. It enforces those boundaries with operators, receipts, and replay — not prompt policies or post-hoc logs.
Ship production value where enterprise risk is highest: external agent actions and durable memory / retrieval basis. Both surfaces emit immutable receipts and bind to operator chains for deterministic replay.
Each surface is a governed product boundary on TraceScript — not a separate language. Surfaces 5 and 6 are runtime proof capabilities shared across the stack.
TrustScript surfaces map to enterprise risk zones. Some application whitepapers span zones (for example Constitutional Agents and Disclosure Firewall) but still run on TrustScript-governed substrate mutation.
What AI believes
Surfaces 2–4 — safe-to-trust before action basis.
What AI says
TrustScript-governed branches — outbound reliance and authority.
What AI does
Surface 1 plus Code Medium — safe-to-act with proof.
Surfaces 5 and 6 — Recursive Continuity Verifier and TraceScript Evidence Export — turn governance decisions into audit-ready artifacts. Operator-bound replay verifies lawful ancestry; RC1 bundles export verification runs for assurance teams.
Each application is a TrustScript-governed branch or surface on TraceScript — read online or download Markdown and PDF editions.
Canonical architecture docs and live OpenAPI routes for the TrustScript runtime layer on RC1.
Full OpenAPI parity on Developers · tenant proof runs on Portal