Content Security Policy
CSP limits which code a browser may execute, buying a second boundary after an output-encoding defect.
Frame the problem
Security starts with a concrete asset, attacker capability and trust crossing.
Why the system fails
The policy allows broad sources, unsafe inline execution, or is treated as the primary XSS fix instead of a backstop.
The important question is not “what is Content Security Policy?” but “which assumption let untrusted data or an over-scoped identity cross rendered document → executable resource?” 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: Start with report-only, then enforce a nonce/hash 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 |
|---|---|---|
| Start with report-only, then enforce a nonce/hash policy · Remove unsafe inline and broad wildcard sources · Keep output encoding as the primary control | Collect and triage CSP violation reports | Contain the affected identity or component, scope impact from audit evidence, and preserve a regression test. |
Key points
- Asset: The browser execution environment.
- Boundary: Rendered document → executable resource
- Primary control: Start with report-only, then enforce a nonce/hash policy
- Detection signal: Collect and triage CSP violation reports
- Always ask what limits damage when the primary control fails.
Follow the attack
Safe conceptual simulation: capability → missing control → crossed boundary → asset impact.
- 1Attacker starts with: An attacker who found an HTML injection or unsafe DOM sink.
- 2The policy allows broad sources, unsafe inline execution, or is treated as the primary XSS fix instead of a backstop.
- 3The weak or missing boundary control is crossed: Rendered document → executable resource
- 4Impact: A surviving injection can execute with the application origin.
- A surviving injection can execute with the application origin.
Defend, detect, recover
One prevention is a single point of security failure. Layer it and make failure observable.
- • Start with report-only, then enforce a nonce/hash policy
- • Remove unsafe inline and broad wildcard sources
- • Keep output encoding as the primary control
- • Collect and triage CSP violation reports
- • 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.