KnownS v0.33.0 suffers from a server‑side request forgery flaw when the POST /api/embedding-models/test endpoint is called. The application opens outbound HTTP connections to URLs supplied by the caller without validating that the target is permitted. An attacker who can trigger this endpoint can force the server to make requests to arbitrary destinations, potentially exposing internal host addresses or cloud metadata services. The vulnerability is a classic CSRF‑like misuse of outbound traffic and maps to CWE‑918, which can lead to information disclosure and style system compromise.

The affected product is the KnownS application. Any installation of KnownS 0.33.0 or earlier that exposes the embedding‑models API to external users is vulnerable; no other versions are listed as affected.

The CVSS score of 6.9 indicates a medium severity that can impact confidentiality and availability. Because the EPSS score is not available, the data does not indicate how frequently this flaw has been exploited in the wild, and it is not listed in CISA’s KEV catalog. The likely attack vector is remote, via an authenticated or unauthenticated client that can send requests to the vulnerable API endpoint. Successful exploitation allows enumeration of internal hosts and metadata services such as cloud instance metadata endpoints – information that can be used to pivot into further network attacks.