Advisory database sources and matching service

Docker Scout is a service that helps developers and security teams build and maintain a secure software supply chain. A key component of this is the ability to assess your software artifacts against a reliable source of vulnerability information. Different tools collect vulnerability information from different sources, and use different methods to identify matches against software artifacts. This can lead to differing results between tools.

To help you understand why different tools can provide different results when assessing software for vulnerabilities, this page explains how the Docker Scout advisory database and CVE-to-package matching service works.

Docker Scout’s advisory database sources

Docker Scout creates and maintains its vulnerability database by ingesting and collating vulnerability data from multiple sources continuously. These sources include many recognizable package repositories and trusted security trackers, including:

Docker Scout correlates the vulnerability data from these advisories with the Software Bill of Materials (SBOM) of container images to detect what vulnerabilities affect an image. The SBOM summarizes the contents of an image, and Docker Scout stores the SBOM in its database.

When there is information about a new vulnerability, Docker Scout correlates the vulnerable package with the SBOMs in the database to identify affected images.

When you enable Docker Scout for your organization, you receive your own instance of the database. The database tracks timestamped metadata about your images that Docker Scout can then match to CVEs. Find more details on how this works in the image analysis page.

Docker Scout image analysis integrates seamlessly with Docker Desktop and Docker Hub, and you can also enable integrations with other systems, see Integrating Docker Scout with other systems.

How Docker Scout makes more precise matches

Many other tools use fuzzy Common Product Enumeration (CPE) matching with wild cards to known vulnerabilities with the versions of software packages they affect. This can return a lot of false positives which you need to triage.

The typical structure of a CPE match looks like this:


For example cpe:*:*:*:calendar:*:*:*:*:*:*:* returns a match on anything with the product name “calendar”. If there is a vulnerability present in an npm package, this CPE match would also return packages and modules for all other languages too.

Instead, Docker Scout matches CVEs to SBOMs using Package URL (PURL) links that are a more precise, universal schema for matching software packages. A PURL link can help you only identify the relevant packages with far less false positives.

Continuing this example, a PURL can match the specific package name to a language and version.


This only matches a node package with the name calendar and the version 12.0.2. For relevant packages, you can specify architectures and operating system versions to make more precise matches.

In summary, Docker Scout’s technique improves matching accuracy and reduces the number of results that turn out to be false-positives.

Package ecosystems supported by Docker Scout

By sourcing vulnerability data from the advisory databases, Docker Scout is able to support analyzing the following package ecosystems:

  • .NET
  • GitHub packages
  • Go
  • Java
  • JavaScript
  • PHP
  • Python
  • RPM
  • Ruby
  • alpm (Arch Linux)
  • apk (Alpine Linux)
  • deb (Debian Linux and derivatives)