GitLab EE EUVD-2026-98208

Requires GitLab Enterprise Edition (EE) running an affected version (17.0 before 19.2.8, 19.3 before 19.3.5, or 19.4 before 19.4.2), and an authenticated user who has been granted a custom role that includes compliance framework administration permissions. Exploitation further depends on the group being configured to use compliance pipelines. Limiting factors: this is not a default configuration (custom roles and compliance framework administration must be deliberately assigned), it is EE-only (not Community Edition), and only limited integrity impact results (C:N/A:N per CVSS), so it is an insider/privileged-account abuse scenario rather than remote unauthenticated compromise.

This is a low-to-moderate priority issue, not a high-urgency one, and the signals are mutually consistent rather than conflicting. The vendor CVSS of 4.3 (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:N) describes a network-reachable but privilege-gated flaw with only limited integrity impact (C:N/A:N), matching CWE-266 (Incorrect Privilege Assignment). Exploitation is gated behind an authenticated user who already holds a custom role with compliance framework administration permissions (PR:L), which sharply narrows the realistic attacker population to insiders or compromised privileged accounts. No EPSS score was supplied, there is no CISA KEV listing, and no public exploit was identified in the provided data, so there is no evidence of active exploitation - treat it as a standard patch-cycle item. The 'Information Disclosure' tag conflicts with the CVSS vector, which explicitly scores C:N (no confidentiality impact) and I:L; the actual weakness is privilege escalation/abuse via compliance pipeline execution, not disclosure, so that tag appears mislabeled.

An authenticated GitLab EE user who has been assigned a custom role carrying compliance framework administration permissions crafts a malicious compliance pipeline and causes it to execute across projects in a group, performing actions beyond what that role was intended to permit. No public exploit code was identified, and the attack requires the attacker to already hold the elevated custom role, so it is primarily an insider or compromised-account abuse path rather than remote unauthenticated takeover.