Secrets Management
A secret should not live everywhere: applications retrieve or receive narrowly scoped values from a controlled system with audit and rotation.
Frame the problem
Security starts with a concrete asset, attacker capability and trust crossing.
Why the system fails
Secrets are committed, copied into images and logs, shared across services or never rotated.
The important question is not “what is Secrets Management?” but “which assumption let untrusted data or an over-scoped identity cross secret manager → authorized workload?” 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 a secret manager or workload identity 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 a secret manager or workload identity · Scope each secret to one purpose and environment · Keep secrets out of source, images, logs and client bundles | Secret scanning and unusual secret reads/uses | Contain the affected identity or component, scope impact from audit evidence, and preserve a regression test. |
Key points
- Asset: API keys, database passwords, signing keys, cloud credentials and encryption keys.
- Boundary: Secret manager → authorized workload
- Primary control: Use a secret manager or workload identity
- Detection signal: Secret scanning and unusual secret reads/uses
- 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 source-code reader, build-log reader or compromised workload.
- 2Secrets are committed, copied into images and logs, shared across services or never rotated.
- 3The weak or missing boundary control is crossed: Secret manager → authorized workload
- 4Impact: Access to every system the secret authorizes, for as long as it remains valid.
- Access to every system the secret authorizes, for as long as it remains valid.
Defend, detect, recover
One prevention is a single point of security failure. Layer it and make failure observable.
- • Use a secret manager or workload identity
- • Scope each secret to one purpose and environment
- • Keep secrets out of source, images, logs and client bundles
- • Secret scanning and unusual secret reads/uses
- • 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.