Loading vulnerability…
CVE-2026-18252 is an inclusion-of-functionality-from-untrusted-control-sphere vulnerability in GitLab Enterprise Edition's Duo Claude AI agent. In affected releases, the agent processes configuration from a user-controlled source, allowing an authenticated user holding the Developer role, under certain conditions, to execute arbitrary commands in a CI context. Affected versions are GitLab EE 18.9 through releases before 19.1.7, 19.2 through releases before 19.2.5, and 19.3 through releases before 19.3.1.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
No public exploits tracked yet. Mallory keeps watching.
No public exploit code observed for this vulnerability.
Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.
Vendor-confirmed product mapping. Mallory continuously reconciles this list against your asset inventory.
13 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.
A GitLab vulnerability included in the referenced FreeBSD security advisory. It is the plugin's CVSS scoring source and is rated High by CVSS v3 (7.3); the content does not map it to a specific listed GitLab issue.
A high-severity arbitrary-command-execution vulnerability in GitLab Enterprise Edition's Duo Claude AI agent. Processing configuration from a user-controlled source could allow an authenticated user with Developer-role permissions to influence and execute commands in a CI environment, potentially exposing pipeline secrets, artifacts, credentials, and connected development infrastructure.
A high-severity arbitrary-command-execution vulnerability in GitLab Enterprise Edition's Duo Claude AI agent. Untrusted, user-controlled configuration could influence CI command execution, potentially exposing CI job secrets, source code, build artifacts, deployment credentials, cloud tokens, and package-registry credentials.
An authenticated command-execution vulnerability in GitLab EE CI contexts. A user with Developer-role permissions could, under certain conditions, execute arbitrary commands by controlling configuration processed by the Claude agent.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.