The reference substrate for the AI OSI Stack and AEIP Protocol.
The AI OSI Kernel is a unified monorepo designed to operationalize artificial intelligence governance as engineered architecture. It consolidates the AI OSI Stack (Layers 0-8), the AI OSI Core decision protocol, and the AI Epistemic Infrastructure Protocol (AEIP) into a single, functional environment.
Unlike traditional governance frameworks that rely on soft policy or retrospective audits, this kernel treats governance as a mandatory system interrupt. It ensures every AI decision is anchored to a civic mandate, bound by temporal constraints, and recorded as immutable evidence.
This monorepo streamlines the previous fragmented repositories into a single source of truth:
/packages/core: The binary conformance logic and schemas for Decision Records (DR)./packages/aeip: The transport SDK for generating signed Epistemic Frames./packages/tower: The backend auditor engine for generating Disclosure Statements (GDS)./apps/dashboard: A mobile-first forensic UI for monitoring real-time maturity and ledger health./specs: Canonical manuscripts for AI OSI Stack v4/v5.
- Binary Conformance: Governance is pass/fail. A decision without a valid mandate or record is a substrate failure.
- Temporal Discipline: AI logic is not permanent. Every record SHALL include a validity limit.
- Epistemic Integrity: Evidence must be captured at the point of origin (substrate/model) and transported without narrative interference.
Phase: Governance Protocol & Reference Implementation This codebase is currently pre-operational. It is intended for inspection, citation, and pilot integration. For the formal research manuscripts, visit aiosi.org.
AI OSI is moving toward a neutral foundation model. We welcome contributions from systems architects, legal counsel, and risk leaders.
© 2025-2026 Daniel P. Madden. Engineered in California.