Executive Summary
Attackers reportedly bypassed GitHub Actions SHA pinning in the Trivy supply-chain hijacking, affecting organizations that rely on GitHub Actions or the scanner. Teams operating AI applications should review third-party action and dependency trust controls.
What Happened
Pinning GitHub Actions to an exact version of their code, known as SHA pinning, has become the go-to defense against software supply chain attacks. Attackers have already found a way around it. In March 2026, the hackers who hijacked Trivy, a widely used open-source security scanner, planted malicious code that looked like it came from GitHub's official checkout tool. They even labeled it a
Affected Products
- GitHub Actions
- Trivy
- GitHub checkout tool
Vulnerability / Attack Technique
Software supply-chain compromise
Technical Mechanism
Attackers planted malicious code during the reported March 2026 Trivy hijacking and labeled it to appear associated with GitHub’s official checkout tool. The summary states that this bypassed the protection normally provided by pinning GitHub Actions to an exact code version, but does not provide further technical details.
AI Security Category
LLM03:2025 Supply Chain
Part of the AI Supply Chain and AI SBOM topic.
Relevant Frameworks
RealGround Analysis
This matters because AI agents and LLM applications commonly depend on open-source scanners, GitHub Actions, and automated build pipelines. A pinned reference alone may not establish trustworthy provenance if the referenced code or release process has been compromised, so teams should map and validate the complete AI software supply chain.
Recommended Controls
- Maintain an inventory of GitHub Actions, open-source libraries, scanners, models, and other AI application dependencies.
- Verify action and dependency provenance using trusted publishers, signed artifacts, and independent integrity checks.
- Review CI/CD workflows for untrusted actions, unexpected code changes, and unauthorized release or repository modifications.
- Require security review and approval for changes to build pipelines and dependencies used by AI agents or LLM applications.
- Monitor third-party security tooling and supply-chain advisories for compromise indicators.
Related RealGround Service
AI Agent SBOM and LLM Mapping · or start with the free AI Agent Security Scanner.
Sources
thehackernews.com: https://thehackernews.com/2026/11/hackers-found-way-around-github-sha.html
Summary and analysis written by RealGround from the source above; last updated 2026-11-18.
