Description

Wyoming before 1.10.2 contains a server-side request forgery vulnerability that allows unauthenticated attackers with network access to force outbound connections to arbitrary targets by supplying a malicious `uri` query parameter to the HTTP API. Attackers can pass arbitrary `tcp://` or `unix://` URIs to affected endpoints including /api/info, /api/speech-to-text, and /api/text-to-speech to override the server-configured backend and redirect connections to attacker-chosen hosts.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "8.3",
  "pubDate": "2026-09-01T19:17:31.180Z",
  "pubdate": "2026-09-01T19:17:31.180Z",
  "executiveSummary": "Wyoming versions prior to 1.10.2 are susceptible to a server-side request forgery (SSRF) vulnerability within its HTTP API.\nThe vulnerability allows unauthenticated attackers with network reachability to the target system to force the server to initiate arbitrary outbound network connections.\nBy manipulating the 'uri' query parameter, an attacker can override the backend configuration to target arbitrary internal or external hosts.\nThis flaw resides in multiple API endpoints, including /api/info, /api/speech-to-text, and /api/text-to-speech.\nThe risk is significant as it facilitates unauthorized interactions with internal services, potential exfiltration of metadata, or service disruption.\nSuccessful exploitation requires no authentication, meaning any actor capable of communicating with the Wyoming HTTP API can leverage this mechanism to interact with unintended endpoints via 'tcp://' or 'unix://' schemes.",
  "technicalDetails": "The root cause of this vulnerability is the improper validation and sanitization of user-supplied input provided via the 'uri' query parameter in the Wyoming HTTP API.\nThe application architecture permits the dynamic specification of backend resources; however, the API fails to enforce a whitelist or validate the structure of the provided URI before passing it to the internal connection handler.\nThis flaw manifests across the /api/info, /api/speech-to-text, and /api/text-to-speech endpoints, which are designed to handle backend resource requests.\nAn unauthenticated attacker can exploit this by crafting HTTP GET or POST requests that inject malicious URI strings into the parameter field. The underlying implementation processes these strings as legitimate connection targets, supporting both 'tcp://' and 'unix://' protocols.\nThe attack flow proceeds as follows: First, the attacker identifies a reachable Wyoming server. Second, the attacker sends an HTTP request to one of the vulnerable endpoints, substituting the expected backend URI with an attacker-defined target (e.g., an internal service port or a Unix domain socket). Third, the Wyoming server, acting as a proxy or client, attempts to open a connection to the attacker-supplied URI. Finally, the server completes the connection, effectively redirecting its outbound traffic to a destination selected by the attacker.\nBecause the server initiates these connections, the attacker can use the service to probe internal network infrastructure that is otherwise protected by firewalls or network segmentation, effectively bypassing perimeter defenses.\nThe impact includes the ability to perform reconnaissance on internal services, exploit local services listening on Unix sockets that lack authentication, or induce the server into unintended interactions with third-party web services.\nThe vulnerability is present in all versions prior to 1.10.2 and requires only network access to the API. No authentication or privileged access is necessary to initiate the malicious request, making this a critical vector for attackers seeking to pivot from a public-facing API into the internal service environment."
}