# CISA KEV Addition: Perforce Puppet Enterprise Report-Processing Deserialization RCE (CVE-2026-115731)

> CVE-2026-115731 is a deserialization-of-untrusted-data flaw in Puppet's report-processing code (Puppet Server / Puppet Enterprise, and the OpenVox fork) that lets a client holding a CA-signed agent certificate run arbitrary code on the server as the puppet user. CISA added it to the Known Exploited Vulnerabilities Catalog on 2026-11-17 based on evidence of active exploitation.

- **Published:** 2026-11-18T03:32:49Z
- **Last reviewed:** 2026-11-18T10:44:03.024Z
- **Canonical:** https://intel.threadlinqs.com/threat/TL-2026-6290
- **ID:** TL-2026-6290
- **Severity:** CRITICAL (CVSS 9.9)
- **Category:** VULNERABILITY
- **Status:** ACTIVE
- **Detections:** 9 (1 shown free below) · **IOCs:** 20
- **CVEs:** CVE-2026-115731, CVE-2026-115732

## Description

CVE-2026-115731 (published by Perforce via CVE.org on 2026-11-03) is a deserialization of untrusted data (CWE-502) in the report-processing code used by Puppet Server, including Perforce Puppet Enterprise. An agent that holds a CA-signed certificate can submit a crafted rich-data report via PUT /puppet/v3/report and obtain code execution on the server as the service user (puppet). The related OpenVox advisory GHSA-ffr6-rwpm-p5wq (reporter: alain-neodyme, published by binford2k on 2026-11-03) documents the root cause: agent-supplied JSON events flow into Puppet::Pops::Serialization::FromDataConverter.convert, where the attacker-controlled __ptype value is parsed by Types::TypeParser and .create(value) is called on the resulting type. This reaches RubyGenerator.new.create_class, where class_eval executes attacker-controlled Ruby.

The advisory's high-level PoC structure is a normal-looking report JSON containing a resource_statuses entry for a Notify resource whose events array holds an object with a __ptype field carrying a serialized Puppet Object type definition. That definition includes a function with a RubyMethod annotation whose Ruby body uses backticks, so the shell command runs during compile/evaluation rather than at parse time. A __pvalue field holds a benign value. Secondary reporting (securityonline.info, dailysecurityreview.com) states a working malicious report that runs a shell command is public in the OpenVox advisory; the working command string was not reproduced in the sources reviewed here.

Authentication prerequisite is a valid CA-signed agent certificate (CVSS Privileges Required: Low). Where the CA autosigns requests without strong verification, an unauthenticated network attacker may obtain such a certificate through the client enrollment path, making the chain effectively unauthenticated. Code execution as the puppet user exposes the CA signing key, which per the advisory could let an attacker issue certificates with serverAuth and clientAuth EKUs and potentially run commands on other hosts in some orchestration setups. dailysecurityreview.com adds that forged certificates can remain valid after patching unless operators rotate trust material (single-source claim).

Severity is reported inconsistently: Red Hat and NVD list CVSS 3.1 9.9 (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H); CVE.org/secondary sources list CVSS 4.0 9.4 (the OpenVox vector is CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H). The CISA alert states the addition is based on evidence of active exploitation but gives no exploitation details, actor, or IOCs; securityonline.info notes its body said no in-the-wild exploitation of 115731 was confirmed at publication on 2026-11-04, while its sidebar lists the KEV addition. Disclosed alongside it was CVE-2026-115732 (CVSS 9.0), a certificate-name authorization bypass in trapperkeeper-authorization (GHSA-q4hm-33rc-572g) allowing node impersonation; securityonline.info reports, citing a Reddit thread (unverified), that this companion bug is exploited in the wild. No threat actor, malware, or network infrastructure has been publicly attributed, so no BeaconBeagle correlation was applicable (no network IOCs exist).

## MITRE ATT&CK

- T1190 Exploit Public-Facing Application
- T1078 Valid Accounts
- T1059 Command and Scripting Interpreter
- T1059.004 Command and Scripting Interpreter: Unix Shell
- T1552.004 Unsecured Credentials: Private Keys
- T1649 Steal or Forge Authentication Certificates
- T1072 Software Deployment Tools
- T1552.001 Unsecured Credentials: Credentials In Files
- T1565.001 Data Manipulation: Stored Data Manipulation
- T1550 Use Alternate Authentication Material

## Free detection (Splunk SPL, Microsoft KQL, Sigma)

**Shell Child Process of Puppet Server JVM Running as puppet User — Post-Exploitation** (`TL-2026-6290-SPL-002`)

Detects a shell (sh, bash, dash, zsh, ksh) spawned by the Puppet Server / OpenVox Server Java process under the puppet service account. In CVE-2026-115731 the attacker-controlled RubyMethod body runs a backtick shell command via class_eval, so a shell child of puppetserver is the primary post-exploitation artifact.

- **Languages:** Splunk SPL, Microsoft KQL, Sigma
- **ATT&CK:** T1059.004, T1059, T1190
- **Data sources:** Sysmon for Linux (index=linux sourcetype=sysmon:linux EventCode=1), DeviceProcessEvents, process_creation linux
- **Severity:** critical
- **Confidence:** high

### Splunk SPL

```spl
index=linux sourcetype="sysmon:linux" EventCode=1 User="puppet" ParentImage="*/java" (ParentCommandLine="*puppetserver*" OR ParentCommandLine="*openvox-server*") (Image="*/sh" OR Image="*/bash" OR Image="*/dash" OR Image="*/zsh" OR Image="*/ksh") earliest=-24h
| stats count, min(_time) AS first_seen, values(CommandLine) AS commands, values(ParentCommandLine) AS parent_cmd BY host, User, Image
| convert ctime(first_seen)
```

### Microsoft KQL

```kql
DeviceProcessEvents
| where Timestamp > ago(24h)
| where InitiatingProcessAccountName =~ 'puppet'
| where InitiatingProcessFileName =~ 'java' and InitiatingProcessCommandLine has_any ('puppetserver', 'openvox-server', 'puppet-server')
| where FileName in~ ('sh', 'bash', 'dash', 'zsh', 'ksh')
| project Timestamp, DeviceName, AccountName, FileName, ProcessCommandLine, InitiatingProcessCommandLine, InitiatingProcessParentFileName
```

### Sigma

```yaml
title: Shell Spawned by Puppet Server JVM as puppet User
id: 2d9f4a61-8c37-4b0e-a5d2-91e6b7c3f840
status: experimental
description: Detects a shell launched as a child of the Puppet Server or OpenVox Server Java process running as the puppet user, the post-exploitation artifact of CVE-2026-115731 report deserialization RCE.
references:
    - https://github.com/OpenVoxProject/openvox/security/advisories/GHSA-ffr6-rwpm-p5wq
    - https://www.cve.org/CVERecord?id=CVE-2026-115731
author: Threadlinqs Team
date: 2026/11/18
tags:
    - attack.execution
    - attack.t1059.004
    - attack.initial-access
    - attack.t1190
    - cve.2026.115731
logsource:
    category: process_creation
    product: linux
detection:
    selection_parent:
        ParentImage|endswith: '/java'
        ParentCommandLine|contains:
            - 'puppetserver'
            - 'openvox-server'
    selection_user:
        User: 'puppet'
    selection_child:
        Image|endswith:
            - '/sh'
            - '/bash'
            - '/dash'
            - '/zsh'
            - '/ksh'
    condition: selection_parent and selection_user and selection_child
falsepositives:
    - Custom Puppet Server hooks or scripts that deliberately shell out from the JVM
level: critical

```

**Known false positives:**

- Custom Puppet Server extensions or hooks that intentionally invoke shell scripts from the JVM
- Administrators debugging the service under the puppet account with an interactive shell

**Tuning notes:** Requires process telemetry with parent command line on the Puppet Server host (Sysmon for Linux, auditd with execve parent enrichment, or EDR). If your build runs the service under a different account, change the User value. Baseline for a week and exclude exact known hook command lines rather than excluding the parent process.

### 8 more detection rules (Splunk SPL, Microsoft KQL, Sigma), unlock with TL Creds

- Puppet Server Log Deserialization Artifacts (Pops Serialization / [indicator]) — Exploitation (locked: unlock with TL Creds)
- Reconnaissance and Persistence Commands Run as puppet Service User — Post-Exploitation (locked: unlock with TL Creds)
- Puppet Server Host Egress and Orchestration Abuse from puppet User — C2 and Lateral Movement (locked: unlock with TL Creds)
- Puppet CA Autosign Activity Enabling Unauthenticated Exploit Chain — Exploitation (locked: unlock with TL Creds)
- Download-and-Execute or Reverse Shell Command Lines Run as puppet User — Post-Exploitation (locked: unlock with TL Creds)
- Puppet Orchestration Abuse and CA-Key Certificate Forging — Lateral Movement (locked: unlock with TL Creds)
- Puppet Server Report Endpoint Deserialization Exploit (CVE-2026-115731) — Exploitation (locked: unlock with TL Creds)
- Puppet CA Private Key Access by Non-JVM Process — Credential Theft and Certificate Forgery (locked: unlock with TL Creds)

Sign up free to unlock the detection rules: https://threadlinqs.com/?modal=signup

## Sources

- [CISA Alert: CISA Adds One Known Exploited Vulnerability to Catalog](https://www.cisa.gov/news-events/alerts/2026/11/17/cisa-adds-one-known-exploited-vulnerability-catalog)
- [CVE Record CVE-2026-115731](https://www.cve.org/CVERecord?id=CVE-2026-115731)
- [OpenVox GHSA-ffr6-rwpm-p5wq: Authenticated RCE via Event Deserialization](https://github.com/OpenVoxProject/openvox/security/advisories/GHSA-ffr6-rwpm-p5wq)
- [Red Hat Customer Portal: CVE-2026-115731](https://access.redhat.com/security/cve/CVE-2026-115731)
- [oss-security: Puppet / OpenVox disclosure](https://www.openwall.com/lists/oss-security/2026/11/03/14)
- [SecurityOnline: Puppet Server RCE CVE-2026-115731 PoC](https://securityonline.info/puppet-server-rce-cve-2026-115731-poc/)
- [Daily Security Review: Public Exploit Released for Puppet Server RCE CVE-2026-115731](https://dailysecurityreview.com/?p=45333)
- [SmartScope: Puppet Server / OpenVox vulnerabilities](https://smartscope.blog/en/blog/puppet-server-vulnerabilities-puppet-server-openvox-allow-server-2026/)
- [CISA Known Exploited Vulnerabilities Catalog](https://www.cisa.gov/known-exploited-vulnerabilities-catalog)
- [CISA BOD 26-04 Implementation Guidance: Prioritizing Security Updates Based on Risk](https://www.cisa.gov/news-events/directives/bod-26-04-implementation-guidance-prioritizing-security-updates-based-risk)

## Full data

The other 8 detection rules (Splunk SPL / Microsoft KQL / Sigma) unlock with TL Creds; a free account includes daily creds: https://threadlinqs.com/?modal=signup
Indicator values unlock with TL Creds, at a lower price on Red and Purple; AI agents can query rules and IOCs through the Threadlinqs MCP server (Purple): https://intel.threadlinqs.com/mcp

Canonical: https://intel.threadlinqs.com/threat/TL-2026-6290
