The flaw in advisor‑backend arises from the use of yaml.Loader to load YAML data from untrusted sources. This unsafe loader permits instantiation of arbitrary Python objects via YAML tags, a well‑known deserialization weakness (CWE‑502). An attacker can embed malicious tags in a payload, causing the application to create and execute arbitrary code when the YAML is parsed.

The vulnerability affects the advisor‑backend application, which typically runs as a pod within a Kubernetes cluster. No specific vendor or version information is provided, but any deployment exposing the YAML parsing endpoint is potentially impacted and could allow an attacker to read database credentials and compromise all tenants served by the backend.

The vulnerability is exploitable by an unauthenticated remote attacker who can send a crafted YAML payload to the backend’s deserialization endpoint. No EPSS metric is available and the issue is not yet listed in CISA KEV, but the potential for remote code execution is high and the attack surface is wide. As a result, the risk to confidentiality, integrity, and availability is significant for any environment exposing the vulnerable endpoint.