Architecture
Runtime Authority
Runtime Authority is the control plane for governed AI execution. It is not a chatbot, not a prompt pack, and not “an AI agent.” It is the durable layer that owns truth for Objectives, Sessions, policy, and Evidence.
The problem
AI Workers can propose useful work and still claim success after failed commands, skip bounds, and leave proof in scrollback. When the model is treated as the system of record, operators inherit that unreliability.
Why prompts are insufficient
A prompt can request careful behavior. It cannot revoke a write grant, stop a loop, or prove a mutation stayed inside an approved path. Policy as natural language alone is a request, not a gate.
Why orchestration is not enough
Tool graphs still leave an open question: who owns Session state, approvals, fail-closed bounds, and inspectable Evidence after the run? Without an authoritative runtime, orchestration is still chat with more steps.
What Runtime Authority owns
It plans Sessions, compiles context, delegates to Workers, enforces policy, collects Evidence, and controls release readiness. Humans stay in the loop wherever mutation requires a decision.
Objectives are not prompts
An Objective states what should be true when the work is done. Runtime Authority turns that into a Session with targets, Runs, Decisions, and Evidence — not a longer system prompt.
Workers are replaceable
Workers are models and tools. They propose. They do not own grants. Swapping a Worker must not redefine policy or invent missing Evidence.
Evidence and Synthetic Releases
If it is not recorded, it is not runtime truth. A Synthetic Release promotes a verified change bundle from Evidence — not from a model saying “done.”
Predator
Predator is the private operator cockpit built on AMOF Runtime Authority. It is the primary operator experience for Sessions, Runs, Decisions, Evidence, and Workforce. It is not part of the public OSS distribution.