A modern application is mostly assembled, not written. Your team writes the business logic; open-source libraries handle logging, JSON parsing, HTTP, date maths, PDF generation and cryptography; a build system compiles it; a container base image provides the operating system layer; a CI pipeline signs and ships it. Each of those pieces is a supplier, and each is a place where someone could slip in something harmful. A software supply chain attack targets one of those suppliers so that the compromise flows downstream to everyone who trusts it.
Four cases that shaped the field
SolarWinds (disclosed December 2020). Attackers compromised the build environment of SolarWinds’ Orion network monitoring product and inserted a backdoor into legitimately signed updates. Thousands of customers installed the tainted update through their normal patching process. The lesson: a valid signature only proves who built the software, not that the build process was clean.
Log4Shell (December 2021). Not an attack on the supply chain but a vulnerability within it. A flaw in the widely used Log4j library affected an enormous range of products, and many organisations spent weeks just discovering where Log4j was present, often buried several dependency layers down. The lesson: you cannot respond quickly to what you cannot see.
Dependency confusion (2021). A security researcher showed that many companies’ build tools, when looking for an internal package name, would also check public registries and prefer a public package with a higher version number. By publishing harmless test packages with the same names as internal ones, he got code to run inside numerous large companies’ build systems. The lesson: package managers resolve names in ways that developers often do not understand.
The xz utils backdoor (March 2024, CVE-2024-3094). A contributor spent around two years building trust in a small but widely used compression project before introducing a carefully hidden backdoor into release archives, targeting SSH on certain Linux distributions. It was caught by a Microsoft engineer who noticed SSH logins taking slightly longer than expected, before it reached most stable distributions. The lesson: overworked, single-maintainer open-source projects are an attractive target for patient social engineering.
The main attack paths
- Typosquatting: publishing malicious packages with names close to popular ones, hoping for a typing mistake.
- Dependency confusion: exploiting resolution order between private and public registries.
- Maintainer account takeover: stealing a maintainer’s registry credentials and publishing a malicious version of a legitimate package.
- Malicious maintainers: either new contributors who gain trust, or existing maintainers who sell or abandon projects.
- Build system compromise: altering code during the build, after review but before signing, as in SolarWinds.
- Install-time scripts: many package ecosystems allow packages to run scripts during installation, which means simply installing a malicious package on a developer laptop or CI runner can compromise it.
What an SBOM is and why it matters
A Software Bill of Materials is a machine-readable inventory of the components in a piece of software: names, versions, suppliers, and how they relate. The two dominant formats are SPDX (from the Linux Foundation, also an ISO standard) and CycloneDX (from OWASP). Build tools and SCA scanners can generate SBOMs automatically.
An SBOM does not prevent anything on its own. Its value is speed and clarity when something goes wrong. When the next Log4Shell lands, an organisation that stores SBOMs for every deployed application can search them and know within an hour which systems are affected. One without SBOMs starts a manual hunt across teams.
Regulators have taken notice. US Executive Order 14028 in May 2021 pushed SBOMs for software sold to the US federal government. In India, CERT-In published technical guidelines on SBOMs in 2024, and sector regulators have been increasing expectations on software inventory for regulated entities. Expect customers in banking, insurance and government projects to ask suppliers for SBOMs more often.
Practical defences for a development team
Pin and lock dependencies. Use lock files so builds install exact, known versions with integrity hashes, rather than “latest compatible”. Review lock file changes in pull requests like any other code.
Use a private registry proxy. Route all package downloads through an internal repository manager that caches approved packages, and configure it so internal package names can never be resolved from public registries. This closes dependency confusion.
Be choosy about new dependencies. Before adding a package, look at its maintenance activity, number of maintainers, download history and whether it is really needed. A twenty-line utility may not justify a new dependency with its own tree of sub-dependencies.
Harden the build pipeline. Treat CI/CD as production infrastructure: least-privilege credentials, short-lived tokens, protected branches, required reviews, and isolated, ephemeral build runners. The SLSA framework (Supply-chain Levels for Software Artifacts) describes progressive levels of build integrity, including provenance records that show where and how an artefact was built.
Sign and verify artefacts. Signing container images and packages, for example with tools from the Sigstore project, and verifying signatures before deployment makes it harder to swap in a tampered artefact.
Run SCA continuously, not just at build time. A dependency that was clean when you shipped can have a CVE published next month. Monitoring deployed SBOMs against new advisories catches this.
The uncomfortable part
Much of the world’s critical software depends on open-source projects maintained by a handful of volunteers. Companies that profit from that code often contribute nothing back. Supply chain security is partly a technical problem and partly a funding and stewardship one. For an individual organisation, the realistic goals are to know what you depend on, limit how much you depend on, control how it enters your builds, and be able to respond fast when one of those dependencies turns out to be compromised.
Written by the RP Strativa Team.
Leave a Reply