Audit Logs for Privileged Actions
Who did what to which resource, when, from where, with what result—and which policy allowed it.
Frame the problem
Security starts with a concrete asset, attacker capability and trust crossing.
Why the system fails
Events omit the principal/resource, are mutable by the actor, or log only successes without authorization decisions.
The important question is not “what is Audit Logs for Privileged Actions?” but “which assumption let untrusted data or an over-scoped identity cross privileged action → tamper-resistant evidence?” Trace the decision at the boundary, then constrain what can happen after the first control fails.
Design the control in layers
Start with the control closest to the interpretation or privilege boundary: Record actor, action, resource, time, source, result and policy Then add a control that reduces blast radius and telemetry that proves the decision was enforced.
The resulting design is not labelled secure. Record the identified controls, the known failure paths, the remaining exposure, and the evidence you would need during an incident.
| Prevent | Detect | Recover |
|---|---|---|
| Record actor, action, resource, time, source, result and policy · Separate audit storage from application write authority · Define retention from investigation needs | Gaps, deletion attempts and impossible action sequences | Contain the affected identity or component, scope impact from audit evidence, and preserve a regression test. |
Key points
- Asset: Accountability and evidence for sensitive operations.
- Boundary: Privileged action → tamper-resistant evidence
- Primary control: Record actor, action, resource, time, source, result and policy
- Detection signal: Gaps, deletion attempts and impossible action sequences
- Always ask what limits damage when the primary control fails.
Boundary control exercise
This lesson uses the shared boundary-control exercise.
Follow the attack
Safe conceptual simulation: capability → missing control → crossed boundary → asset impact.
- 1Attacker starts with: A privileged user or compromised identity attempting misuse or repudiation.
- 2Events omit the principal/resource, are mutable by the actor, or log only successes without authorization decisions.
- 3The weak or missing boundary control is crossed: Privileged action → tamper-resistant evidence
- 4Impact: Misuse cannot be detected, scoped or attributed.
- Misuse cannot be detected, scoped or attributed.
Defend, detect, recover
One prevention is a single point of security failure. Layer it and make failure observable.
- • Record actor, action, resource, time, source, result and policy
- • Separate audit storage from application write authority
- • Define retention from investigation needs
- • Gaps, deletion attempts and impossible action sequences
- • Contain the affected identity or component.
- • Scope access from audit evidence.
- • Fix the boundary and add a regression test.
- • Misconfiguration and new access paths can bypass the intended control.
- • A privileged insider or compromised control plane may still reach the asset.