#access-control

Claude and ChatGPT now log into your app as your customers. OneHuman decides what they can see and do. Join our community: https://discord.gg/9QX9jTwBq

Deterministic authorization + tamper-evident audit for AI agents, humans, and workloads — one principal type, safety invariants machine-checked by an SMT solver (cvc5), every decision independently verifiable offline. Pure Rust.

Specification for governance-first context memory in autonomous AI agent systems. Defines the Mandate Gate: a pre-work mandate must exist before an agent session is valid.

Empirical evidence that OWASP AISVS C9.2 has no accountability axis: an action an AI agent cannot be answerable for reaches human-owned execution only through a false input. By K. Lunina.