
An open standard for agentic products
An engineering specification for building LLM-agent products that can be trusted: boundaries of autonomy, verifiable quality gates, human control over money and irreversible actions. Published in full, with five reference implementations assembled into one open platform.
