Northwatch

Northwatch

Puppet Server report deserialization flaw is now on CISA's exploited list

CVE-2026-115731 lets any holder of a CA-signed Puppet agent certificate run code on Puppet Server. CISA added it to KEV on 17 November. Who is exposed, why autosigning matters, and what to do.

Published

On 17 November 2026, CISA added CVE-2026-115731 to its Known Exploited Vulnerabilities (KEV) catalog. The flaw is in Puppet’s report-processing code, as used by Puppet Server, and it lets a client with a valid agent certificate execute arbitrary code on the server. CISA lists it as “Perforce Puppet Enterprise Deserialization of Untrusted Data Vulnerability” and has set a remediation due date of 1 December 2026 for agencies covered by its directives.

Puppet Server is the control point for every node it manages. Code execution there is not a single-host problem. Anyone who controls the server can, in principle, influence the configuration that is pushed to the whole fleet. That is why this one deserves attention even if you rarely think about your configuration management tier as an attack surface.

What the flaw is

The NVD record, published on 3 November 2026 and now marked as analysed, describes the issue like this: deserialization of untrusted data in the report-processing code of Puppet allows “a client that holds any CA-signed agent certificate to execute arbitrary code on the Puppet Server as its service user by submitting a report containing a crafted rich-data type via PUT request.”

Two weakness classes are attached to the KEV entry: CWE-502 (deserialization of untrusted data) and CWE-94 (code injection).

In plain terms:

  1. Puppet agents send run reports back to the server over an authenticated HTTPS request.
  2. The server parses those reports, and the parsing accepts rich data types.
  3. A malicious report can carry a crafted rich-data value that, when the server deserialises it, results in code running on the server.
  4. That code runs as the Puppet Server service user.

The only credential required is a certificate signed by the Puppet certificate authority. It does not need to be an administrator’s certificate, and it does not need to belong to any particular node. Any agent certificate will do.

Why autosigning turns this into an unauthenticated bug

On its own, “requires a valid agent certificate” sounds like a reasonable barrier. Many attackers do not have one. The catch is autosigning.

Puppet’s documentation explains that before an agent can retrieve its configuration catalog, it needs a certificate signed by the Puppet CA. By default those certificate signing requests (CSRs) must be signed manually by an administrator, either with the puppetserver ca command or on the Node requests page in the Puppet Enterprise console. To speed up onboarding, the CA can be configured to sign some CSRs automatically, and the documentation warns that autosigning “changes the nature of your deployment’s security.”

The NVD and CISA descriptions spell out the consequence for this bug: where the certificate authority autosigns requests without strong verification, an unauthenticated network attacker can obtain a valid certificate and then exploit the flaw. In that configuration the attack chain is simply: reach the Puppet Server, request a certificate, get it signed automatically, send a malicious report.

So the realistic exposure question has two parts:

  • Can untrusted networks reach your Puppet Server’s agent-facing port?
  • Does your CA sign certificate requests without a strong check of who is asking?

If both answers are yes, treat this as an unauthenticated remote code execution issue on a critical system.

Severity and affected versions

The CVSS 4.0 score provided by Puppet (via security@puppet.com) and shown on NVD is 9.4, rated critical. The vector is network-reachable, low complexity, no user interaction, and low privileges required. The low-privilege requirement is the agent certificate. Impact is rated high for confidentiality, integrity, and availability, both on the vulnerable system and on subsequent systems. That last part reflects the fleet-wide reach of a compromised Puppet Server.

NVD’s configuration data lists these affected ranges:

Product Affected First fixed version implied by NVD
Puppet 8.0.0 up to, but not including, 8.22.2 8.22.2
Puppet 9.0.0 up to, but not including, 9.1.3 9.1.3
Puppet Enterprise 2023.8.0 up to, but not including, 2023.8.13 2023.8.13
Puppet Enterprise 2025.0.0 up to, but not including, 2025.12.1 2025.12.1
Puppet Enterprise 2026.0.0 see vendor advisory

Check the Perforce advisory for the authoritative fix guidance for your release stream, especially for Puppet Enterprise 2026.0.0, where NVD lists the single version without an end range.

What “known exploited” means here

KEV inclusion means CISA has evidence that the vulnerability is being exploited. It does not, by itself, tell you who is exploiting it or how widely. The KEV entry lists known ransomware campaign use as “Unknown”, and no public technical write-up of in-the-wild activity was referenced in the catalog entry at the time of writing.

CISA’s required action points to its Binding Operational Directive 26-04, “Prioritizing Security Updates Based on Risk”, issued on 10 June 2026. The directive is binding on US federal civilian agencies only, but the KEV due date is a useful signal for everyone else: CISA considers this worth fixing within two weeks of listing.

Practical takeaways

1. Find your Puppet Servers and their versions. Include the primary server, any compile servers, and test or lab instances that were stood up and forgotten. Compare each version against the table above.

2. Patch first. Move to a fixed release in your stream (for open source and Core: 8.22.2 or later on 8.x, 9.1.3 or later on 9.x; for Puppet Enterprise, follow the Perforce advisory). Remember that the vulnerable code runs on the server, so the server is what needs updating.

3. Review autosigning today, even before you patch. If your CA autosigns requests with no meaningful verification, that is the setting that removes the authentication barrier. Switch to manual signing or a policy that genuinely verifies the requester until you are patched. Puppet’s documentation describes the different autosigning approaches and their security impact.

4. Restrict network access to Puppet Server. Agents need to reach it, but the internet usually does not. Limit the agent-facing port to the networks where your managed nodes live.

5. Audit the certificates your CA has already signed. List the certificates the CA has issued and look for names you do not recognise, especially recent ones. A certificate that should not exist is a strong signal that someone has been probing or has already got in. Revoke anything you cannot account for.

6. Look for signs of compromise on the server. Because the payload runs as the Puppet Server service user, review that account’s activity: unexpected child processes, new files in the server’s code and configuration directories, and unexpected changes to Puppet code or Hiera data that would push to nodes. If you find something, assume the managed fleet may also be affected and scope the investigation accordingly.

7. Treat configuration management as tier zero. Puppet Server, like other orchestration and deployment systems, can change every host it manages. It belongs in the same protection tier as your identity providers: tight network exposure, strict enrolment, and prompt patching.

The bigger lesson

This is a familiar pattern: a deserialization bug in a component that trusts its own clients too much, combined with an enrolment shortcut that hands those clients out freely. Neither part is unusual on its own. Together they turn a “low privileges required” issue into something an unauthenticated attacker can use against the most powerful server in many estates.

If you run Puppet, the fix is straightforward: patch the server, tighten autosigning, and check who already holds a certificate.

Sources

  1. CISA Known Exploited Vulnerabilities Catalog: CVE-2026-115731
  2. NVD: CVE-2026-115731
  3. Perforce advisory: Report deserialization code execution
  4. Puppet Core documentation: Autosigning certificate requests
  5. CISA BOD 26-04: Prioritizing Security Updates Based on Risk