Modern software is assembled more than written. A typical application contains hundreds of open-source dependencies, built by automated pipelines, delivered through package registries and update channels. Each link is an opportunity for attackers. Compromised build systems, malicious packages, hijacked maintainer accounts and backdoored updates have caused some of the most significant incidents of recent years.
The main attack paths
- Malicious or compromised dependencies: typosquatted package names, takeovers of abandoned packages, or a malicious update pushed by a compromised maintainer account.
- Vulnerable dependencies: a widely used library with a critical flaw that suddenly affects thousands of products.
- Build system compromise: attackers alter the build process so that legitimate source code produces a backdoored binary.
- Update channel compromise: signed, trusted updates that carry malicious code.
- Leaked secrets in repositories and pipelines, giving attackers access to publish.
SBOMs: knowing what's inside
A software bill of materials (SBOM) is a machine-readable list of the components in a piece of software: names, versions, suppliers and relationships. The two main formats are SPDX and CycloneDX.
SBOMs help you answer, within hours instead of weeks, the question every major vulnerability raises: are we affected, and where? In India, CERT-In has published technical guidelines on SBOMs and related bills of materials, and sector regulators increasingly expect them from software suppliers.
For SBOMs to be useful:
- generate them automatically in the build, not by hand;
- store them and link them to deployed versions;
- feed them into vulnerability management, matching against vulnerability databases continuously;
- use VEX (vulnerability exploitability exchange) statements to record which vulnerabilities actually affect the product.
Provenance and SLSA
Knowing what is inside is not enough; you need confidence in how it was built. SLSA (Supply-chain Levels for Software Artifacts) is a framework of increasing assurance levels for build integrity: scripted builds, a hosted build service, signed provenance describing how the artefact was built, and hardened, isolated builds.
Signing
Signing proves who produced an artefact and that it hasn't changed since:
- sign commits, container images and release artefacts;
- verify signatures at deployment, for example with admission controllers that reject unsigned images;
- tools such as Sigstore make signing and verification easier by tying signatures to identities and recording them in a transparency log;
- protect signing keys in HSMs or managed key services, and plan for post-quantum signing algorithms over time. See crypto-agility.
Harden your own pipeline
- Pin dependencies to versions and verify hashes; use a curated internal registry or proxy.
- Run software composition analysis and secret scanning on every change.
- Protect the CI/CD system like production: MFA, least privilege, short-lived credentials (workload identity federation instead of stored keys), and isolated build runners.
- Require code review and branch protection for main branches and pipeline definitions.
- Monitor for unexpected changes to build scripts and pipeline configuration.
What to ask software vendors
Add these to due diligence for critical software. See third-party risk management:
- Do you produce an SBOM for each release, in SPDX or CycloneDX, and will you share it?
- How quickly do you assess and fix vulnerabilities in dependencies, and do you publish VEX or advisories?
- How do you secure your build pipeline? Do you produce signed provenance (for example SLSA)?
- Are your releases and updates signed, and how can we verify them?
- How do you protect signing keys and developer accounts?
- Have you had a supply-chain incident, and how did you handle it?
- What is your secure development framework, and is it independently assessed?
AI adds new supply-chain risks
Models, datasets and AI libraries are now part of the software supply chain: pre-trained models with hidden behaviours, poisoned datasets, and malicious model files that execute code when loaded. Apply the same disciplines: provenance, scanning, pinned versions and trusted sources. The OWASP Top 10 for LLM Applications lists supply chain as a major risk.
Learn more
Secure software development and supply-chain security are covered in the free CISSP and CCSP tracks, and AI supply-chain risks in our AI Security & Governance course.
Written and reviewed by the DigiTrust Asia editorial team. How we research, review and correct articles.