The situation
Credentials and compliance rules don’t stop at one cloud. With work on both AWS and GCP, secrets needed a single, controlled way of being handled, and security checks needed to happen before code ships, not after.
What I built
- Set up secrets management with Vault, AWS Secrets Manager and GCP Secret Manager to meet compliance requirements, so credentials don’t live in code or pipeline config.
- Put scanners in the pipeline as gates. In my own projects, Trivy, hadolint and tflint run on every push and a failing scan stops the build. Snyk and SonarQube are the dependency and code-analysis tools I use in the same position.
How it flows
- Commit
- Code analysis
- Dependency scan
- Image scan
- Gate
- Deploy
Decisions I made and why
Fail the build, don’t file a report
A finding that blocks a merge gets fixed that day. A finding in a report gets fixed eventually.
Block on what can be fixed
Gating on critical issues that have a fix available keeps the signal high. Teams take the gate seriously because it rarely cries wolf.
Secrets come from a manager at runtime
Pulling credentials from Vault or a cloud secret manager means rotation doesn’t require a code change, and nothing sensitive sits in Git.
What changed
Secrets moved into managed stores that help meet compliance requirements, and on my own projects a problem now surfaces in the pull request instead of in production.
What I’d do next
Sign container images and verify signatures at deploy time, so only images built by the pipeline can run.
Stack Vault, AWS Secrets Manager, GCP Secret Manager, Trivy, hadolint, tflint, Snyk, SonarQube
This website uses the same idea on a small scale: Trivy and hadolint run on every push, and a failing scan blocks the deploy.
Details about State Farm’s systems are kept general on purpose. The approach, the tools and the results are from my work there.