Skip to content
AIGP

RFC-021: AIGP-SGL — Symbolic Governance Language for Agentic and Autonomous AI — 6. Layer Model

AIGP Specification › RFC-021: AIGP-SGL — Symbolic Governance Language for Agentic and Autonomous AI › 6. Layer Model

← 5. Non-Goals · Section index · 7. Core Concepts →

6. Layer Model

AIGP-SGL operates between human intent and machine enforcement.

Human Intent Layer
  Natural language, doctrine, law, ethics, mission, organizational rules


AIGP-SGL Layer
  Symbolic governance notation and typed governance statements


Canonical AST Layer
  Machine-readable abstract syntax tree


Compiled Enforcement Layer
  Rego, Cedar, XACML, SpiceDB checks, finite-state machines, runtime monitors


Runtime Decision Layer
  permit, deny, inhibit, modify, escalate, degrade, stop, return


D-DNA Evidence Layer
  signed temporal evidence of authority, intent, constraint, decision, action, consequence

The source of runtime enforcement is not the human-readable explanation.

The source of runtime enforcement is:

Governance Intent
+ AIGP-SGL statement
+ canonical AST
+ compiled constraint
+ runtime context
+ verified authority
+ D-DNA evidence binding

← 5. Non-Goals · Section index · 7. Core Concepts →