Docker and ActiveState extend secure defaults to the packages developers install

Docker has announced a partnership with ActiveState, a company that, according to the article itself, has been securing open-source software since 1997, to extend its trusted content standard beyond container images to the open-source packages that applications install at build time. The announcement, published on November 5, 2026 by Vishrut Iyengar, security product marketing lead at Docker, has a clear commercial angle, so its figures and promises should be read as vendor claims.

The starting point is Docker Hardened Images (DHI). According to the article, they are open source under the Apache 2.0 license and make up a catalog of more than 4,000 minimal images, built from source and continuously maintained by Docker. The catalog also includes customizable Helm charts and more than 65,000 hardened system packages. Docker says that for most services, adopting them is as simple as changing the FROM line in the Dockerfile, and that the images can get close to zero known CVEs, cutting remediation work and scanner noise.

The core argument is that above the base image sits another layer of the supply chain: the libraries that pip, npm or Maven pull from public registries, where trust is assumed and rarely verified. On top of that, AI coding agents also install these packages. The text cites an April article by Mark Lechner, CISO, which argues that a compromised dependency pulled in by an agent has the same blast radius as a compromised developer machine.

The article mentions several recent incidents. It says that when the TeamPCP campaign weaponized a widely used security scanner in March, DHI were not affected thanks to their controlled build pipeline and dependency cooldown periods. It also cites the axios compromise earlier this year, in which two backdoored versions of a library downloaded 83 million times a week were in circulation for about three hours. The lesson it draws is that many of these attacks die out within hours, but a build that resolves dependencies directly from the public registry, without a lockfile or cooldown, is exposed to versions published just an hour earlier; and that the slow failure is the vulnerability that surfaces months later with no one outside the team responsible for fixing it.

On the guarantees at each layer, Docker says every DHI includes a signed SBOM, SLSA Build Level 3 provenance and full CVE visibility with no suppressions. Paid tiers add a 7-day SLA for critical and high CVEs, FIPS and STIG variants, and secure customization.

ActiveState, for its part, contributes its Curated Catalog: a private, policy-governed repository of the open-source components that security teams approve and that developers and AI coding assistants consume. According to the text, ActiveState builds each component from the original source code in a SLSA Level 3 environment, with full provenance and complete SBOMs, rather than using prebuilt binaries from a public registry. When an upstream fix appears, it automatically rebuilds and publishes the component, with an SLA of 5 business days for critical CVEs and 10 for high-severity ones.

In practice, the artifact repository teams already use, such as Artifactory or Nexus, is configured to point to the Curated Catalog as its upstream, just as it points to PyPI today. Developers keep their pip and npm workflows, and agents get the governed version of the dependencies. The article clarifies that this does not replace the pinned digests, cooldown periods or sandboxed agents Lechner recommends, but gives those controls a governed source to point to.

Commercially, the deal is reciprocal: Docker now resells ActiveState's Curated Catalog alongside DHI, and ActiveState resells DHI to its customers. As a result, both the hardened image and the packages built on top of it can be purchased through Docker.

The underlying message is that AI agents are speeding up software creation, dependency selection and build launches at a pace that traditional review processes were not designed for, and that this speed should not translate into accepting more unchecked supply chain risk. Docker argues that the goal is not to add another friction-inducing checkpoint, but to make the secure path the easiest one, with strong defaults and a foundation maintained by a team accountable for it.

In assessment, the proposal is consistent with the trend toward curated internal repositories, verifiable provenance and remediation SLAs, but the text offers no technical details of the integration, catalog pricing or independent data to verify the CVE figures or response times cited.