AI coding assistants are writing more code than ever, and every line they generate pulls in open-source dependencies that nobody is manually reviewing. Chainguard, the supply-chain security company, just shipped four products aimed squarely at that gap.
At its Assemble 2026 conference in New York on March 19, the company rolled out Chainguard Actions, Agent Skills, a unified Repository, and Commercial Builds. Together, they represent a bet that securing the software supply chain (the full trail of code, packages, and tools that go into a deployed application) is now an AI problem, not just a human one.
What Actually Shipped
Chainguard Actions are pre-secured CI/CD workflows, starting with GitHub Actions. The system automatically ingests popular third-party workflows, scans them against security rules, fixes unsafe patterns like excessive permissions or dependency confusion risks, and republishes hardened versions. Each comes with a software bill of materials (SBOM) and provenance attestations so teams can verify exactly what they're running. Beta launched March 17.
CEO Dan Lorenc put it bluntly: "CI/CD pipelines power modern software delivery, but the privileged workflows remain one of the least secured layers."
Chainguard Agent Skills is the more unusual launch. As AI coding agents like Cursor, Claude Code, and GitHub Copilot gain the ability to use external tools and plugins, those plugins become attack surface. Chainguard's catalog ingests skills from community registries, reviews them against security rules, and hardens them using its own reconciliation agents before publishing with full audit trails.
Chainguard Repository bundles everything (containers, libraries, OS packages, CI/CD workflows, and agent skills) into a single endpoint with policy enforcement baked in. Think of it as one place to pull all your dependencies where the security review already happened.
Commercial Builds is a partner program where software vendors ship Chainguard-hardened container images. Early partners include GitLab, Elastic, Grafana Labs, F5 NGINX, and Azul.
The Actual Problem Being Solved
Here's the tension: AI coding tools help developers write and ship faster, but they also pull in dramatically more open-source packages. More packages means more potential vulnerabilities, and human security teams can't review at that speed. Lorenc framed it as a scaling problem: "The challenge is no longer just generating code, but using AI and secure pipelines to scale review and ensure software is safe to ship."
Chainguard is also expanding beyond container images into Python, Java, and JavaScript library artifacts, which tracks with where AI-generated code actually lives. Most AI coding assistants aren't producing Docker containers directly. They're writing Python scripts and JavaScript functions that import dozens of packages.
Who This Is For
Most individual developers and small teams won't interact with Chainguard directly. This is enterprise infrastructure, the kind of thing a platform engineering team deploys so that everyone else can move fast without thinking about supply chain attacks.
But the Agent Skills piece is worth watching for anyone building with AI coding agents. Right now, the plugin and tool ecosystems around AI agents are mostly unaudited. Someone has to start vetting those, and Chainguard is positioning itself as that layer.
The broader signal here: as AI writes more of our code, the security industry is scrambling to build automated review systems that can keep pace. Patching after deployment was already losing the race. With AI accelerating the pipeline, it's not even close.