IBM MQ CVE-2026-12667

Exploitation requires: (1) the victim to be running an affected IBM MQ .NET client (not the queue manager itself - this is a client-side flaw); (2) an authenticated attacker (PR:L) with sufficient authority to place messages onto a queue that the vulnerable .NET client consumes; and (3) the crafted message to include a malicious XML external entity inside its RFH2 folder, which the client parses without disabling external-entity resolution. Limiting factors: the attacker must already hold valid MQ credentials with queue-write access, the target must actually deploy and use the .NET client (Java/C clients and the server are not indicated as affected), and the confirmed impact is bounded to file disclosure from the client host and limited (not full) denial of service.

This is a genuine but moderate-priority issue, well below its 7.1 CVSS would casually suggest for a Log4Shell-era enterprise-messaging product. The CVSS vector (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L) and SSVC (Exploitation: none, Automatable: no, Technical Impact: partial) are consistent with each other and point to a targeted, not opportunistic, threat. Critically, this is a CLIENT-side XXE: the affected component is the IBM MQ .NET client that parses RFH2 message folders, so the party at risk is the application host running that client, not the queue manager. An authenticated attacker (PR:L) who can place messages onto a queue the .NET client consumes can read files from the client host (C:H) or cause limited denial of service (A:L, e.g. entity-expansion). No EPSS score was provided, there is no CISA KEV listing (Exploitation: none), and no public exploit was supplied - so this is 'no public exploit identified at time of analysis.' The main signals-in-tension are the network attack vector (AV:N) versus the reality that the attacker must already be an authenticated MQ participant able to write to a consumed queue, which narrows the realistic attacker population considerably.

An attacker with credentials sufficient to put messages on a queue crafts an MQ message whose RFH2 folder contains a malicious XML external entity referencing a local file (or a resource that triggers entity expansion). When a vulnerable IBM MQ .NET client consumes and parses that RFH2 folder, the XXE resolves, disclosing local file contents back to the attacker or causing a limited denial of service on the client. No public exploit code was identified; exploitation requires authenticated write access to a queue that a target .NET client reads.