---
title: "Policy decision point (PDP): definition | KYA glossary"
url: https://knowyouragent.ch/en/knowledge/glossary/policy-decision-point
lang: en
updated: 2026-10-12
license: © Innopulse Consulting GmbH
---

# Policy decision point (PDP)

A policy decision point (PDP) is the component that decides, based on rules, whether a requested action is allowed; a policy enforcement point (PEP) enforces that decision.

The separation of decision and enforcement comes from access control and is described, for example, in the XACML standard and in NIST's zero trust architecture. The PEP sits in the data flow, for example in a gateway, and queries the PDP. The PDP evaluates rules, attributes and context and answers with allow or deny. Widespread tools are Open Policy Agent and Cedar.

For AI agents, this separation is central: the decision about an action must not lie in the language model, which can be influenced by content, but in a separate component with fixed rules.

## Example

An agent wants to trigger a payment of CHF 900. The gateway (PEP) queries the PDP, which checks the mandate: limit CHF 500. The PDP denies and signs this decision; the gateway executes nothing.

## How it differs

The PDP decides, the PEP enforces, and the mandate provides the rules for the specific instruction. A [decision token](https://knowyouragent.ch/en/knowledge/glossary/decision-token) is the PDP's signed result, which can be verified later.

## In practice

- Place the decision about agents' actions in a separate component, never in the language model.
- Have the PDP sign every decision so it remains traceable later.
- Keep rules versioned so it is clear which version applied to a decision.

## Role in KYA

In the KYA architecture, the decision service takes on the role of the PDP. KYA verifies the signed decision and whether it matches the mandate and the executed action.

Related check: [Scope of the action](https://knowyouragent.ch/en/method/checks) (`intent_scope`). Is the action within the mandate's action type, amount limit, currency and counterparty?

[Verify now: open a matching example](https://knowyouragent.ch/en/verify?beispiel=abgelehnt)

## Frequently asked questions

### What is a policy decision point?

The component that decides, based on rules and context, whether an action is allowed. It makes the decision but does not enforce it itself.

### What is a policy enforcement point?

The component in the data flow, such as a gateway, that queries the PDP and enforces its decision, executing or blocking the action.

### Why does the decision not belong in the language model?

Because language models can be influenced by content, for example through prompt injection. A separate component with fixed rules remains unaffected.

### Which tools exist for PDPs?

Widespread are Open Policy Agent with the Rego policy language and Cedar. For [agent mandates](https://knowyouragent.ch/en/knowledge/glossary/agent-mandate), a check of the mandate's limits is added.

### How does the PDP relate to KYA?

The PDP corresponds to the decision service. Its signed result, the decision, is one of the pieces of evidence KYA verifies.

## Sources

- [NIST SP 800-207: Zero Trust Architecture](https://csrc.nist.gov/pubs/sp/800/207/final)
- [Open Policy Agent](https://www.openpolicyagent.org/)

## Related terms

- [Decision token](https://knowyouragent.ch/en/knowledge/glossary/decision-token)
- [Zero trust](https://knowyouragent.ch/en/knowledge/glossary/zero-trust)
- [Agent mandate](https://knowyouragent.ch/en/knowledge/glossary/agent-mandate)
- [Human in the loop](https://knowyouragent.ch/en/knowledge/glossary/human-in-the-loop)

---
Quelle: https://knowyouragent.ch/en/knowledge/glossary/policy-decision-point · KYA – Know Your Agent