Action Risk Tier Policy

Data artifactSafety & SecuritySafety, Security & Governancearc:ActionRiskTierPolicy

An externalized configuration classifying each agent action as autonomous, approval-required or prohibited, with unknown actions defaulting to prohibited.

Responsibility. Assigns every agent action a risk tier.

Also known as: Action policies, Action classification policy, Decision authority boundaries, Graduated autonomy tiers, Graduated response policy, Decision boundary map

Direct neighbourhood (hover for relationship types)

Relationships

configures structural

is read by dependency

Autonomy Scope Adjuster Ch10.2

Design guidance

MUST default unknown actions to PROHIBITED (fail secure).
SHOULD be stored in a configuration file or database so security teams can modify policies without code changes.
SHOULD allow runtime context (user, circumstances) to change an action's decision.
SHOULD tier actions by risk and reversibility: routine reversible actions autonomous, irreversible or high-impact actions (freezing accounts, blocking transactions, contacting customers, deleting repositories, committing to protected branches) approval-required.
SHOULD partition the operational space along multiple dimensions simultaneously (data classification, transaction value, reversibility) rather than a single approval threshold.

Classification

Patterns: Fail secureThree-tier action classificationRisk-based decision boundariesReversibility-based tiering
Risks mitigated: Accidental privilege escalation from unclassified new capabilities

Sources

Ch9.1: T. Nguyen, "Output Filtering and Content Moderation," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 9.1. ISBN: 9798244538229.
Ch10.2: T. Nguyen, "Proactive Agents," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 10.2. ISBN: 9798244538229.
Ch10.4: T. Nguyen, "Human-in-the-Loop," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 10.4. ISBN: 9798244538229.
Ch10.5: T. Nguyen, "Human-over-the-Loop," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 10.5. ISBN: 9798244538229.
Ref9.04: "Safety Guardrails Implementation for Agent Systems," unpublished reference note (04-Safety-Guardrails-Implementation.md), Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam supplementary materials, 2026. unpublished note