Skip to main content
Policies are the rules that govern what AI agents can and cannot do. In TealTiger v1.3, policies are authored by governance teams, signed into bundles, and enforced deterministically at runtime.

What Is a Policy?

A policy is a rule that maps a condition to an action with an automation level:
Every policy evaluation produces a deterministic Decision — same input + same policy = same decision, every time.

Policy Structure


Automation Levels

Each rule specifies how the decision should be automated:

FREEZE Rules

FREEZE rules are a special class of policy that cannot be overridden:
Properties:
  • Evaluated FIRST (before all other rules)
  • Cannot be modified, disabled, or removed by application code
  • Persist across policy hot-swaps
  • Log FREEZE_TAMPER_ATTEMPT on any modification attempt
See FREEZE Rules for details.

Policy Bundles

In enterprise deployments, policies are packaged into signed bundles:
Bundles include: policy rules, FREEZE rules, NHI configuration, cost ceilings, and module settings. See Governance at Scale.

Quick Start Examples

Zero-Config (OWASP Pack)

Covers all 10 OWASP Agentic Security risks with production-ready defaults.

Custom Rules


Policy Evaluation Order

  1. FREEZE rules — Checked first, short-circuit on match
  2. NHI scope — Agent must have permission for the action
  3. Temporal constraints — Session TTL, cooldowns, time-of-day
  4. Policy rules — Matched in order, first match wins
  5. Default — If no rule matches, action is ALLOW (configurable)

Next Steps

Policy Authoring Guide

Write custom policies with conditions and matchers

Reason Codes

All available reason codes and their meanings

OWASP Policy Pack

Pre-built zero-config governance

Configuration Reference

Full engine configuration options