---
title: "Authorization Is a Runtime Decision"
description: "See how an agent request becomes an allowed or denied tool action through identity context, policy checks, scoped credentials, and an auditable decision trail."
---

# Authorization Is a Runtime Decision

**Type:** [How it works](https://yomu.fyi/visual-explainers/category/how-it-works.md)

![A left-to-right numbered flow shows an agent request carrying agent, user, and organization context, passing through a policy gate, receiving scoped credentials, calling a tool, and producing an audit record or denial.](https://yomufyi.s3.us-east-1.amazonaws.com/visual-explainers/5ed3d658-c516-4765-aceb-4dfe300bad2c/image-1f74c576-5752-4578-a775-f66d559c9c4b.webp)

> See how an agent request becomes an allowed or denied tool action through identity context, policy checks, scoped credentials, and an auditable decision trail.

## Explanation

Authorization decides whether a known principal may perform a specific action on a specific resource; authentication only establishes identity. This explainer follows an agent request from agent, user, and organization context through a policy decision, short-lived or user-scoped credentials, tool execution, and audit logging. It also shows where consent, tenant boundaries, read-only limits, human review, and revocation fit into the runtime path.

## FAQ

### Why is an agent’s identity alone not enough to authorize a tool call?

The gateway example evaluates agent, user, and organization context together because customer-facing agents may act across delegated accounts and tenant boundaries. A recognized identity still needs a runtime decision about the requested tool action and scope.

### Where should credentials and permissions be limited in an agent workflow?

Use short-lived agent-only tokens or user-scoped credentials rather than treating the agent as permanently privileged. The supplied examples also describe delegated OAuth access with read-only enforcement, token lifetimes, and revocation.

### What should happen when an action needs more than an automated policy decision?

A policy check can route sensitive cases to a durable human pause instead of automatic approval. The fraud-review example records recommendations, human verdicts, audit events, and mandate receipts so the decision remains traceable.

**Tags:** [AI Agents](https://yomu.fyi/topic/ai-agents.md), [Auditability](https://yomu.fyi/topic/auditability.md), [Authentication](https://yomu.fyi/topic/authentication.md), [Authorization](https://yomu.fyi/topic/authorization.md), [OAuth2](https://yomu.fyi/topic/oauth2.md)

## Related reading

### [Auth0 Agent Gateway: The Identity Control Plane for Customer-Facing AI](https://yomu.fyi/post/auth0-agent-gateway-the-identity-control-plane-for-customer-facing-ai.md)
- Company: [Auth0](https://yomu.fyi/company/auth0.md)
- Author: Andrés Aguiar
- Published: Sep 18, 2026
- Type: Announcement

Auth0 Agent Gateway is announced in beta as a Customer Identity and Access Management (CIAM) control plane for governing product-native, customer-facing AI agents in multi-tenant SaaS applications. It addresses a trust model involving organizations, delegated user access, customer consent, independently owned connected accounts, and cross-tenant data boundaries, rather than workforce-only identity. Positioned between an agent and its tools, the gateway evaluates each call using agent, user, and organization context; it uses Organizations, Agents as Principals, Token Vault, token exchange, tenant logs, and administrative controls. The described capabilities include short-lived agent-only tokens, user-scoped credentials, runtime tool authorization, audit logging with optional SIEM streaming, and an agent kill switch. Auth0 is seeking product and engineering teams to join the beta.