RabbitMQ EUVD-2026-87156

Exploitation requires that the rabbitmq_auth_backend_ldap plugin is enabled AND configured with a user_dn_pattern that contains the ${username} placeholder (the injectable substitution point). Additionally, the target directory layout must be such that the attacker's injected DN suffix resolves to a useful alternate entry, and the attacker must possess a password that is valid for that resulting DN. These are non-default, deployment-specific prerequisites: standard RabbitMQ installations without the LDAP backend, or LDAP configurations that do not template the username directly into the DN, are not exploitable. The combination of required directory conditions and a valid password is reflected in the CVSS AC:H and AT:P metrics.

This is a genuine but narrowly-conditioned LDAP DN injection (CWE-90), not the high-priority issue a raw score might imply. The provided CVSS 4.0 vector (6.3, AV:N/AC:H/AT:P/PR:N/UI:N/VC:N/VI:L/VA:N) is internally consistent with the description: network-reachable but high attack complexity (AC:H) and an attack requirement (AT:P) reflect that exploitation depends on a specific directory layout in which the injected DN suffix resolves usefully. Critically, the attacker must also supply a password valid for the resulting DN, which sharply constrains real-world abuse - this is closer to authorization-boundary manipulation by someone with partial LDAP-valid credentials than to unauthenticated compromise. No EPSS score, no CISA KEV listing, and no public exploit are present in the input, so this is no public exploit identified at time of analysis. Confidentiality and availability impact are rated none; only limited integrity (VI:L) applies. The signals do not conflict on severity, but the input DOES contain a material data conflict on remediation: the advisory body lists fixed versions 3.13.15 / 4.0.20 / 4.1.11, while structured metadata lists 3.13.18 / 4.0.23 / 4.1.14 (both agree only on 4.2.9 and 4.3.3). Per the advisory itself, no fixed-version assertion for the 3.13/4.0/4.1 lines is certifiable until a curator resolves this - patch guidance for those lines must be verified with the vendor.

An attacker who can reach the RabbitMQ authentication endpoint and holds a password that will validate against some alternate directory entry submits a username containing RFC 4514 DN metacharacters (e.g. a comma-delimited suffix). Because fill/2 substitutes ${username} into user_dn_pattern without DN escaping, the crafted suffix rewrites the bind DN so the bind resolves to a different, attacker-chosen directory entry. No public proof-of-concept is available, and success additionally requires a directory layout where the injected suffix is meaningful.