Agent Image Security Updates
How agent images are patched, and what to do with container scanner findings
If you scan the Monte Carlo agent image with a container scanner (for example, Wiz), this page explains how the image is patched and what to do with the findings.
Where agent images are published
| Deployment | Registry | Tags |
|---|---|---|
| AWS Lambda (default CloudFormation template or Terraform module) | Monte Carlo's Amazon ECR repository | The version number, e.g. 1.4.20, and latest. A new deployment starts on latest. |
| Azure, GCP Cloud Run and other platforms | Docker Hub montecarlodata/agent | The version plus a platform suffix, e.g. 1.4.20-azure, 1.4.20-cloudrun, 1.4.20-aws-proxied or 1.4.20-lambda |
AWS Lambda deployments run the Amazon ECR image, not the -lambda image on Docker Hub.
What changed in each version is listed in the agent's release notes on GitHub. New versions are released often.
How images are patched
- Python dependencies are upgraded in agent releases to pick up security fixes. When a release upgrades a dependency, its release notes name it.
- Operating system packages come from the image's base operating system. Every release is built fresh from the current base image and applies the latest security updates that the base operating system has published at that time.
- Images aren't rebuilt between releases. To pick up newer operating system fixes, upgrade to the latest version.
- Images on Docker Hub carry an SBOM (software bill of materials) and build provenance attestations, which you can inspect with your own tools. The Amazon ECR images used by AWS Lambda don't, because AWS Lambda can't run images that include them.
Some findings in the operating system layer have no fix in the base operating system yet. They stay in every image until the base operating system publishes a fix, and then disappear at the next release. Upgrading can't remove them sooner.
What to do with scanner findings
- Upgrade the agent to the latest version and scan again. Many findings are fixed in a newer release. See Upgrading.
- Check what's being scanned. Scanners also report on old or unused agent deployments and on cached images. Make sure the finding is for an agent that's running.
- If findings remain, ask Bayes to open a support ticket. Include the image tag, the scanner's resource name, and an export of the findings. Monte Carlo's security team reviews which remaining findings can be fixed and which can't be reached by the agent.
For how Monte Carlo prioritizes and fixes vulnerabilities, see the Vulnerability Management Policy in the Trust Center.
Updated about 16 hours ago
