Intermediate · Systems

Agent Authorization

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. Deciding and enforcing what an authenticated agent is allowed to do, at what scale, and under which conditions.

Technical Definition

Deciding and enforcing what an authenticated agent is allowed to do, at what scale, and under which conditions.

How it works

Agent authorization enforces least privilege for non-human actors. It differs from ordinary service authorization in three ways: authority is usually delegated from a human principal and must be traceable to them; agents can act thousands of times per minute, so limits (rate, spend, blast radius) matter as much as permissions; and agents are steerable by untrusted input, so authorization must be enforced outside the model. Practical implementations grant narrow, short-lived, action-level permissions and require an explicit escalation path for anything irreversible.

Related Concepts

• Context as a Security Boundary — The design principle that an agent's context window is a trust boundary: what enters it defines what the agent can be persuaded to do.
• Agent Authorization Protocol (AAP) — A protocol that determines which actions an authenticated agent is permitted to perform, on whose behalf, and within what limits.
• Agent Capabilities — The concrete set of actions an agent can perform — the tools, APIs, and side effects available to it.
• Capability-Based Agent Authorization — Authorizing agents by handing them unforgeable, narrowly scoped capability tokens rather than checking roles at call time.