Define the software you actually ship
A supply-chain program begins with an inventory of source repositories, direct and transitive dependencies, build services, package registries, and release destinations. Map who can modify each component and which credentials can publish code or artifacts.
Treat that inventory as living operational data. An SBOM can improve visibility into components, but an outdated or incomplete SBOM does not prove a release is safe. Pair inventory with ownership, update cadence, and a process for handling new advisories.
Protect the source and build identities
Require strong authentication for maintainers and review changes to protected branches. Limit build-service tokens to the repositories, workflows, and artifact stores they actually need. Avoid long-lived credentials in job logs, configuration files, and container images.
Separating build and release authority reduces the chance that one compromised developer account can silently ship a malicious artifact. Record who approved releases and retain trustworthy logs of privileged pipeline changes.
Make builds repeatable and observable
Pin trusted build tools and dependencies where practical, verify integrity of downloaded components, and isolate jobs that process untrusted contributions. Treat a pull request from an external contributor as untrusted code that should not automatically receive production secrets.
Generate signed provenance and build metadata when your ecosystem supports them. Reproducibility helps teams compare expected and delivered artifacts, while traceability shortens investigations when a dependency or build worker is compromised.
Use several kinds of validation
Dependency scanning, static analysis, secret detection, license review, and dynamic application testing find different classes of problems. Run checks early, but reserve stronger enforcement for material risks so developers do not learn to ignore a wall of low-value warnings.
Define an exception process with an owner, expiration date, and compensating controls. A vulnerability finding is a starting point; exploitation prerequisites, deployed exposure, and the existence of mitigations affect remediation decisions.
Prepare for a compromised dependency
Practice locating affected applications from your component inventory, rebuilding clean artifacts, revoking credentials, and notifying service owners. Record which artifacts were published in the suspect period and where they were deployed.
The best maturity indicator is not a dashboard full of green checks. It is the ability to answer what shipped, who changed it, how it was built, where it is running, and how quickly it can be replaced.
Practical takeaways
- Maintain dependency and release inventories.
- Use least-privilege identities in build pipelines.
- Validate artifacts and record build provenance.
- Rehearse the response to a compromised package.
Official references and further reading
Publisher pages may change; verify current versions and licensing.