RabbitMQ CVE-2026-57221

Exploitation requires a valid RabbitMQ credential with permission to connect to at least one virtual host - the authorization bypass occurs only after this initial authentication gate is cleared. The flaw is exclusive to the AMQP 0-9-1 protocol interface; deployments that expose only AMQP 1.0, MQTT, or STOMP endpoints to untrusted users are not vulnerable through those protocols. No configure, write, or read permission grants are needed beyond basic vhost connectivity, and no administrative or management-plugin access is required. The attack is bounded to information disclosure - enumeration of queue and exchange names, message counts, and consumer counts - with no write or availability impact possible through this vector. Exploitation is further limited to the scope of the connected virtual host; cross-vhost enumeration is not enabled by this flaw alone.

The CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L/UI:N/VC:L/VI:N/VA:N/SC:L/SI:N/SA:N) reflects a network-accessible flaw exploitable with only low-privilege authenticated access, requiring no attack complexity or user interaction. The provided base score of 5.3 is consistent with a limited information-disclosure impact. EPSS data was not provided in the source intelligence, preventing probabilistic exploitation likelihood assessment. The vulnerability is not in CISA KEV and no POC has been identified, placing exploitation status at the lowest confirmed tier. Practical risk centers on internal reconnaissance: an attacker holding any valid RabbitMQ credential can map broker topology - queue names, exchange names, message depths, consumer counts - useful intelligence for targeting high-value queues or planning lateral movement in microservice environments. The broad version range across four concurrent release branches suggests a significant unpatched population in production. A spurious 'Microsoft' tag appears in source metadata and has no bearing on risk assessment. The SC:L signal in the provided CVSS 4.0 vector implies downstream system impact is possible if disclosed topology enables follow-on attacks against dependent services; this is speculative rather than directly evidenced.

An attacker holding any valid RabbitMQ credential - such as a compromised low-privilege service account or a shared application credential - connects to a target virtual host over the standard AMQP 0-9-1 interface and sends a series of passive queue.declare frames for arbitrary or guessed queue names. Because the broker performs no authorization check on these operations, it returns confirmation, message counts, and consumer counts for each existing queue, allowing the attacker to enumerate the full broker topology without triggering permission errors. This reconnaissance output can then be used to identify high-throughput queues carrying sensitive payloads or to time follow-on attacks against consumer services. No public exploit code has been identified at time of analysis.