CI/CD Security
Pull-request code, build runners, secrets and deployment authority meet in CI; trust must change across fork, branch and environment boundaries.
Frame the problem
Security starts with a concrete asset, attacker capability and trust crossing.
Why the system fails
Untrusted pull-request code runs with secrets, persistent runners retain state, or one pipeline can directly change production.
The important question is not “what is CI/CD Security?” but “which assumption let untrusted data or an over-scoped identity cross untrusted code change → privileged build/deploy workflow?” 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: Separate untrusted tests from trusted release jobs 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 |
|---|---|---|
| Separate untrusted tests from trusted release jobs · Use ephemeral runners and short-lived scoped identity · Require reviewed promotion of immutable artifacts | Unexpected secret access, network calls and release provenance | Contain the affected identity or component, scope impact from audit evidence, and preserve a regression test. |
Key points
- Asset: Build secrets, source integrity, artifacts and deployment control.
- Boundary: Untrusted code change → privileged build/deploy workflow
- Primary control: Separate untrusted tests from trusted release jobs
- Detection signal: Unexpected secret access, network calls and release provenance
- 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 contributor, compromised dependency or stolen CI identity.
- 2Untrusted pull-request code runs with secrets, persistent runners retain state, or one pipeline can directly change production.
- 3The weak or missing boundary control is crossed: Untrusted code change → privileged build/deploy workflow
- 4Impact: Credential theft, artifact tampering or production compromise.
- Credential theft, artifact tampering or production compromise.
Defend, detect, recover
One prevention is a single point of security failure. Layer it and make failure observable.
- • Separate untrusted tests from trusted release jobs
- • Use ephemeral runners and short-lived scoped identity
- • Require reviewed promotion of immutable artifacts
- • Unexpected secret access, network calls and release provenance
- • 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.