Actively Exploited Puppet Enterprise Flaw Puts Fleet Configuration Servers At Risk
A critical remote code execution vulnerability in Perforce Puppet Enterprise, tracked as CVE-2026-115731, is being exploited in the wild against the very servers organisations use to centrally configure and manage their endpoint fleets. The flaw is a deserialization of untrusted data weakness in the Puppet Server component that ingests the reports agents send back after each configuration run. Affected products span Puppet Enterprise 2023.8.0 before 2023.8.13, 2025.0.0 before 2025.12.1 and 2026.0.0 before 2026.0.1, as well as Puppet Core 8.x before 8.22.2 and 9.0.0 before 9.1.3. OpenVox, the community-maintained open-source Puppet fork, is affected too, with OpenVox Agent versions before 9.3.0 and 8.32.0 and OpenVox Server versions before 9.2.0 and 8.18.0 all vulnerable. Perforce published fixes on 3 November alongside a patch for CVE-2026-115732, a certificate name authorization bypass in the same product line.
Puppet runs on a client-server model in which agents on each managed machine periodically check in with a central Puppet Server, pull their configuration and report back, with both sides authenticating through certificates issued by the Puppet Server's own certificate authority. CVE-2026-115731 abuses that reporting path: a client holding any certificate signed by the Puppet CA can submit a report containing a crafted rich-data type via a PUT request, causing the server to execute arbitrary code with the privileges of the Puppet Server service account. The requirement for a signed certificate looks like a barrier, but it largely evaporates where the CA signs requests without strong verification. With autosigning enabled, which is common in larger deployments, an attacker on the network can simply request a valid certificate through the client enrollment endpoint and turn the issue into an unauthenticated attack chain. The OpenVox project's advisory, published alongside its fixes, includes a proof-of-concept malicious report in the form of a JSON event that runs a shell command on the server. Puppet Server listens on TCP port 8140 by default, so any instance reachable from untrusted networks is directly in scope.
The strategic weight here comes from what the Puppet Server controls rather than from the bug itself. Code execution as the puppet service user grants access to the CA signing key, which the OpenVox maintainers note would let an attacker mint certificates carrying EKU critical, serverAuth and clientAuth attributes, in turn enabling direct command execution on other hosts where certain kinds of orchestration are configured. A compromise therefore does not stop at one server: it reaches the configuration code that defines how every managed endpoint is built, offering a path to confidentiality loss through sensitive data held on the server, integrity loss through altered configuration, and availability loss through service disruption. It also means remediation and incident response are different problems, since a server showing signs of compromise leaves the certificate authority itself untrustworthy. Exploitation is confirmed rather than theoretical: CISA added CVE-2026-115731 to its Known Exploited Vulnerabilities catalog on 17 November 2026 and set a 1 December 2026 remediation deadline for US federal civilian agencies. Details of the in-the-wild attacks have not been made public. FIRST EPSS currently places the 30-day probability of exploitation at 0.7% for both CVE-2026-115731 and CVE-2026-115732, a figure that sits in tension with the KEV listing and should be read as a statistical forecast rather than a contradiction of observed attacks. For any organisation running Puppet with autosigning or an internet-reachable Puppet Server, the realistic worst case is loss of control over the entire managed estate.
Attack Surface
System Management Service, Infrastructure, Server OS
Tactics
Initial Access, Execution, Credential Access, Lateral Movement, Impact
Techniques
- T1190 – Exploit Public-Facing Application
- T1059 – Command and Scripting Interpreter
- T1552.004 – Unsecured Credentials: Private Keys
- T1649 – Steal or Forge Authentication Certificates
- T1072 – Software Deployment Tools
SuperPRO's Threat Countermeasures Procedures
- Upgrade Puppet Enterprise or Puppet Advanced to 2026.0.1, 2025.12.1 or 2023.8.13, and Puppet Core to 9.1.3 or 8.22.2, then restart the Puppet Server service so the fixed report-processing code is loaded.
- For OpenVox deployments, upgrade OpenVox Server to 9.2.0 or 8.18.0 and OpenVox Agent to 9.3.0 or 8.32.0, which carry the fix for the same deserialization flaw in the report endpoint.
- Where upgrading cannot be scheduled immediately, review the autosign configuration (autosign settings in puppet.conf and any autosign.conf or autosign script) and disable blanket autosigning so certificates are not issued to unverified nodes via the client enrollment endpoint.
- Tighten Puppet Server authorization rules in auth.conf to restrict which authenticated clients may issue PUT requests to the report submission endpoint, rather than permitting any CA-signed certificate.
- Restrict inbound access to TCP/8140 at the firewall so only known agent subnets and compiler hosts can reach the Puppet Server, and remove any exposure of that port to the internet or untrusted segments.
- Audit the list of signed certificates on the CA for node names that do not match inventory, and hunt on the Puppet Server for unexpected processes, newly written files or outbound connections running under the Puppet Server service user.
- If compromise indicators are found, treat the Puppet CA as compromised: rebuild the certificate authority and reissue agent certificates, since patching alone does not revoke certificates an attacker could have signed with the stolen CA key.