Create an exception using the VEX

もくじ

Vulnerability Exploitability eXchange (VEX) is a standard format for documenting vulnerabilities in the context of a software package or product. Docker Scout supports VEX documents to create exceptions for vulnerabilities in images.

メモ

You can also create exceptions using the Docker Scout Dashboard or Docker Desktop. The GUI provides a user-friendly interface for creating exceptions, and it's easy to manage exceptions for multiple images. It also lets you create exceptions for multiple images, or your entire organization, all at once. For more information, see Create an exception using the GUI.

Prerequisites

To create exceptions using OpenVEX documents, you need:

  • The latest version of Docker Desktop or the Docker Scout CLI plugin
  • The vexctl command line tool.
  • The containerd image store must be enabled
  • Write permissions to the registry repository where the image is stored

Introduction to VEX

The VEX standard is defined by a working group by the United States Cybersecurity and Infrastructure Security Agency (CISA). At the core of VEX are exploitability assessments. These assessments describe the status of a given CVE for a product. The possible vulnerability statuses in VEX are:

  • Not affected: No remediation is required regarding this vulnerability.
  • Affected: Actions are recommended to remediate or address this vulnerability.
  • Fixed: These product versions contain a fix for the vulnerability.
  • Under investigation: It is not yet known whether these product versions are affected by the vulnerability. An update will be provided in a later release.

There are multiple implementations and formats of VEX. Docker Scout supports the OpenVex implementation. Regardless of the specific implementation, the core idea is the same: to provide a framework for describing the impact of vulnerabilities. Key components of VEX regardless of implementation includes:

VEX document
A type of security advisory for storing VEX statements. The format of the document depends on the specific implementation.
VEX statement
Describes the status of a vulnerability in a product, whether it's exploitable, and whether there are ways to remediate the issue.
Justification and impact
Depending on the vulnerability status, statements include a justification or impact statement describing why a product is or isn't affected.
Action statements
Describe how to remediate or mitigate the vulnerability.

vexctl example

The following example command creates a VEX document stating that:

  • The software product described by this VEX document is the Docker image example/app:v1
  • The image contains the npm package express@4.17.1
  • The npm package is affected by a known vulnerability: CVE-2022-24999
  • The image is unaffected by the CVE, because the vulnerable code is never executed in containers that run this image
$ vexctl create \
  --author="author@example.com" \
  --product="pkg:docker/example/app@v1" \
  --subcomponents="pkg:npm/express@4.17.1" \
  --vuln="CVE-2022-24999" \
  --status="not_affected" \
  --justification="vulnerable_code_not_in_execute_path" \
  --file="CVE-2022-24999.vex.json"

Here's a description of the options in this example:

--author
The email of the author of the VEX document.
--product
Package URL (PURL) of the Docker image. A PURL is an identifier for the image in a standardized format, defined in the PURL specification.

Docker image PURL strings begin with a pkg:docker type prefix, followed by the image repository and version (the image tag or SHA256 digest). Unlike image tags, where the version is specified like example/app:v1, in PURL the image repository and version are separated by an @.

--subcomponents
PURL of the vulnerable package in the image. In this example, the vulnerability exists in an npm package, so the --subcomponents PURL is the identifier for the npm package name and version (pkg:npm/express@4.17.1).

If the same vulnerability exists in multiple packages, vexctl lets you specify the --subcomponents flag multiple times for a single create command.

You can also omit --subcomponents, in which case the VEX statement applies to the entire image.

--vuln
ID of the CVE that the VEX statement addresses.
--status
This is the status label of the vulnerability. This describes the relationship between the software (--product) and the CVE (--vuln). The possible values for the status label in OpenVEX are:
  • not_affected
  • affected
  • fixed
  • under_investigation

In this example, the VEX statement asserts that the Docker image is not_affected by the vulnerability. The not_affected status is the only status that results in CVE suppression, where the CVE is filtered out of the analysis results. The other statuses are useful for documentation purposes, but they do not work for creating exceptions. For more information about all the possible status labels, see Status Labels in the OpenVEX specification.

--justification
Justifies the not_affected status label, informing why the product is not affected by the vulnerability. In this case, the justification given is vulnerable_code_not_in_execute_path, signalling that the vulnerability can't be executed as used by the product.

In OpenVEX, status justifications can have one of the five possible values:

  • component_not_present
  • vulnerable_code_not_present
  • vulnerable_code_not_in_execute_path
  • vulnerable_code_cannot_be_controlled_by_adversary
  • inline_mitigations_already_exist

For more information about these values and their definitions, see Status Justifications in the OpenVEX specification.

--file
Filename of the VEX document output

Example JSON document

Here's the OpenVEX JSON generated by this command:

{
  "@context": "https://openvex.dev/ns/v0.2.0",
  "@id": "https://openvex.dev/docs/public/vex-749f79b50f5f2f0f07747c2de9f1239b37c2bda663579f87a35e5f0fdfc13de5",
  "author": "author@example.com",
  "timestamp": "2024-05-27T13:20:22.395824+02:00",
  "version": 1,
  "statements": [
    {
      "vulnerability": {
        "name": "CVE-2022-24999"
      },
      "timestamp": "2024-05-27T13:20:22.395829+02:00",
      "products": [
        {
          "@id": "pkg:docker/example/app@v1",
          "subcomponents": [
            {
              "@id": "pkg:npm/express@4.17.1"
            }
          ]
        }
      ],
      "status": "not_affected",
      "justification": "vulnerable_code_not_in_execute_path"
    }
  ]
}

Understanding how VEX documents are supposed to be structured can be a bit of a mouthful. The OpenVEX specification describes the format and all the possible properties of documents and statements. For the full details, refer to the specification to learn more about the available fields and how to create a well-formed OpenVEX document.

To learn more about the available flags and syntax of the vexctl CLI tool and how to install it, refer to the vexctl GitHub repository.

Verifying VEX documents

To test whether the VEX documents you create are well-formed and produce the expected results, use the docker scout cves command with the --vex-location flag to apply a VEX document to a local image analysis using the CLI.

The following command invokes a local image analysis that incorporates all VEX documents in the specified location, using the --vex-location flag. In this example, the CLI is instructed to look for VEX documents in the current working directory.

$ docker scout cves <IMAGE> --vex-location .

The output of the docker scout cves command displays the results with any VEX statements found in under the --vex-location location factored into the results. For example, CVEs assigned a status of not_affected are filtered out from the results. If the output doesn't seem to take the VEX statements into account, that's an indication that the VEX documents might be invalid in some way.

Things to look out for include:

  • The PURL of a Docker image must begin with pkg:docker/ followed by the image name.
  • In a Docker image PURL, the image name and version is separated by @. An image named example/myapp:1.0 has the following PURL: pkg:docker/example/myapp@1.0.
  • Remember to specify an author (it's a mandatory field in OpenVEX)
  • The OpenVEX specification describes how and when to use justification, impact_statement, and other fields in the VEX documents. Specifying these in an incorrect way results in an invalid document. Make sure your VEX documents comply with the OpenVEX specification.

Attach VEX documents to images

When you've created a VEX document, you can attach it to your image in the following ways:

You can't remove a VEX document from an image once it's been added. For documents attached as attestations, you can create a new VEX document and attach it to the image again. Doing so will overwrite the previous VEX document (but it won't remove the attestation). For images where the VEX document has been embedded in the image's filesystem, you need to rebuild the image to change the VEX document.

Attestation

To attach VEX documents as an attestation, you can use the docker scout attestation add CLI command. Using attestations is the recommended option for attaching exceptions to images when using VEX.

You can attach attestations to images that have already been pushed to a registry. You don't need to build or push the image again. Additionally, having the exceptions attached to the image as attestations means consumers can inspect the exceptions for an image, directly from the registry.

To attach an attestation to an image:

  1. Build the image and push it to a registry.

    $ docker build --provenance=true --sbom=true --tag <IMAGE> --push .

  2. Attach the exception to the image as an attestation.

    $ docker scout attestation add \
  --file <cve-id>.vex.json \
  --predicate-type https://openvex.dev/ns/v0.2.0 \
  <IMAGE>

    The options for this command are:

    • --file: the location and filename of the VEX document
    • --predicate-type: the in-toto predicateType for OpenVEX

Image filesystem

Embedding VEX documents directly on the image filesystem is a good option if you know the exceptions ahead of time, before you build the image. And it's relatively easy; just COPY the VEX document to the image in your Dockerfile.

The downside with this approach is that you can't change or update the exception later. Image layers are immutable, so anything you put in the image's filesystem is there forever. Attaching the document as an attestation provides better flexibility.

メモ

VEX documents embedded in the image filesystem are not considered for images that have attestations. If your image has any attestations, Docker Scout will only look for exceptions in the attestations, and not in the image filesystem.

If you want to use the VEX document embedded in the image filesystem, you must remove the attestation from the image. Note that provenance attestations may be added automatically for images. To ensure that no attestations are added to the image, you can explicitly disable both SBOM and provenance attestations using the --provenance=false and --sbom=false flags when building the image.

To embed a VEX document on the image filesystem, COPY the file into the image as part of the image build. The following example shows how to copy all VEX documents under .vex/ in the build context, to /var/lib/db in the image.

# syntax=docker/dockerfile:1

FROM alpine
COPY .vex/* /var/lib/db/

The filename of the VEX document must match the *.vex.json glob pattern. It doesn't matter where on the image's filesystem you store the file.

Note that the copied files must be part of the filesystem of the final image, For multi-stage builds, the documents must persist in the final stage.