Security-Safe Logging
Logs are production data stores: they should contain decision evidence, not passwords, tokens, full payment details or unnecessary personal data.
Frame the problem
Security starts with a concrete asset, attacker capability and trust crossing.
Why the system fails
Request bodies, headers or tool output are logged wholesale; secrets spread into every sink and retention tier.
The important question is not “what is Security-Safe Logging?” but “which assumption let untrusted data or an over-scoped identity cross application event → long-lived broadly queried telemetry?” 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: Use an allowlisted structured event schema 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 |
|---|---|---|
| Use an allowlisted structured event schema · Redact at the source and minimize personal data · Protect log integrity and access | Secret/PII scans and malformed event fields | Contain the affected identity or component, scope impact from audit evidence, and preserve a regression test. |
Key points
- Asset: Credentials, personal data and the integrity of operational evidence.
- Boundary: Application event → long-lived broadly queried telemetry
- Primary control: Use an allowlisted structured event schema
- Detection signal: Secret/PII scans and malformed event fields
- 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: Anyone with log access or a path that injects misleading log content.
- 2Request bodies, headers or tool output are logged wholesale; secrets spread into every sink and retention tier.
- 3The weak or missing boundary control is crossed: Application event → long-lived broadly queried telemetry
- 4Impact: Secondary data breach, token reuse or corrupted investigations.
- Secondary data breach, token reuse or corrupted investigations.
Defend, detect, recover
One prevention is a single point of security failure. Layer it and make failure observable.
- • Use an allowlisted structured event schema
- • Redact at the source and minimize personal data
- • Protect log integrity and access
- • Secret/PII scans and malformed event fields
- • 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.