IBM MQ Agent EUVD-2026-71259

Exploitation requires a valid authenticated IBM MQ Console session cookie, placing this squarely in the authenticated insider or compromised-credential threat model. The attack targets the AI Agent feature specifically; only users with access to submit LLM agent requests can trigger the condition. No additional privilege escalation, special configuration, or non-default deployment mode is required beyond basic console authentication. Limiting factors: the attack does not affect core IBM MQ message-broker functionality, only the AI Agent feature; impact is bounded to availability of that specific feature; and the requirement for authenticated access prevents exploitation by unauthenticated external attackers.

The CVSS 6.5 score with vector AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H correctly reflects the moderate-to-high risk profile for availability. The authentication requirement (PR:L) is the primary mitigation - exploitation is limited to users who already hold valid IBM MQ Console session cookies, meaning this is an insider or compromised-credential threat. Attack complexity is low once access is obtained; no special configuration or race condition is needed. Impact is purely availability-focused (A:H): the AI Agent feature becomes fully unavailable to other users, though core IBM MQ messaging functionality is presumably unaffected. EPSS data is not provided, and the vulnerability is not listed in CISA KEV, indicating no confirmed active exploitation. No POC is publicly available. Real-world priority is moderate: organizations where the AI Agent feature is business-critical should patch promptly; those treating it as supplemental tooling face lower urgency. The main concern is a disgruntled insider or a compromised low-privileged account being used to deny service.

An authenticated IBM MQ Console user - such as a disgruntled employee or an attacker who has compromised a low-privileged account - submits a series of concurrently crafted, maximally complex LLM agent requests. Within seconds, all available agent workers are occupied processing these expensive requests, some for over ten minutes each. All other MQ Console users attempting to use the AI Agent feature receive errors or timeouts for the duration of the attack.