Implement an agent policy enforcement point

A PEP must intercept every path that can create a protected effect.

01 / The operating case

What changes in a real workflow?

An incident agent can isolate host H-7 through a tool wrapper and through a direct endpoint API.

Trace the authority boundary
  1. 01Authority origin
  2. 02Task and grant
  3. 03Current decision
  4. 04Observed outcome

02 / Decision contract

What the executor must check

Normalize verb and immutable host ID, request a decision immediately before execution, bind the result to those values, and block both routes on DENY or unavailable authority.

InputPrincipal · task · action · target
DecisionGrant ancestry · policy · current state
ResultEnforce · record · verify outcome

03 / Failure and evidence

What goes wrong, and what can be proven?

Failure mode

The wrapper enforces policy but a direct API call uses the same cloud credential without interception.

Evidence to retain

Compare PEP decisions with endpoint API invocations; investigate every unpaired operation.

04 / Canonical scope

Why this reference stands alone

Policy-decision-point implementation computes a verdict; this guide owns where and how to enforce it.

Illustrative implementation artifact

PEP boundary test

for path in [toolWrapper, directApi, queuedWorker]:
  attempt delete(customerId) without task grant
  assert path returns DENY
  assert customer state unchanged

Primary references

Read the underlying material

Architecture discussion

Choose one consequential action and make its boundary explicit.

Request a Conversation