Severity:

HIGH RISK

(8.3)

Server-Side Request Forgery (CWE-918) in Wyoming API

CVE-2026-8712

Deferred

Vulnerability type

Category:

Server-Side Request Forgery

CWE:

CWE-918

Impact:

Unauthorized outbound network connections and potential access to internal services

Severity:

HIGH RISK

(8.3)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

Wyoming contains a server-side request forgery vulnerability in its HTTP API that allows unauthenticated network attackers to influence outbound connections through a malicious uri query parameter.

When affected endpoints accept attacker-controlled tcp:// or unix:// URIs, the server may connect to arbitrary attacker-selected targets instead of the configured backend.

Affected context

The vulnerability occurs in Wyoming deployments that expose the affected HTTP API endpoints and allow the uri query parameter to override the configured backend. Risk is higher when the service can reach internal networks, local Unix sockets, cloud metadata services, or other privileged network destinations.

Exploit preconditions

  • Attack surface: HTTP API endpoints including /api/info, /api/speech-to-text, and /api/text-to-speech
  • Required access: Network access to the affected API; authentication is not required according to the vulnerability description
  • Trigger point: Processing a request containing a malicious uri query parameter
  • Why exploitation is possible: The server accepts attacker-controlled tcp:// or unix:// destinations and uses them to establish outbound connections, overriding the configured backend

Why this happens

The underlying flaw is insufficient validation and authorization of a user-controlled URI used for backend selection. Instead of treating the configured backend as trusted application state, the HTTP API allows request data to influence connection destinations without adequately restricting schemes, hosts, ports, or local socket paths.

This behavior maps to CWE-918: Server-Side Request Forgery, because attacker-supplied input causes the server to make network or local inter-process connections on the attacker's behalf.

Potential impact

  • An attacker may probe internal services that are not directly exposed
  • The server could connect to cloud metadata or administrative endpoints
  • Sensitive responses may be disclosed through API behavior or timing
  • The attacker could reach privileged services accessible from the Wyoming host
  • Local Unix sockets could be abused to interact with unintended processes
  • Internal network reconnaissance and follow-on attacks could become easier

General mitigation guidance

Mitigation generally focuses on removing attacker control over connection destinations and enforcing an explicit outbound trust policy.

Common defensive patterns include:

  • Ignore or reject client-supplied backend URI overrides unless they are explicitly required
  • Permit only approved schemes, hosts, ports, and Unix socket paths through an allowlist
  • Resolve and validate destinations before connecting, including protections against DNS and address-based bypasses
  • Block access to loopback, link-local, private, and metadata-service ranges when they are not required
  • Apply network egress controls and isolate the service from sensitive internal segments
  • Run the service with minimal filesystem, socket, and network privileges
  • Return generic errors that do not disclose internal connection details

How teams detect issues like this

Detection approaches may include:

  • Static analysis and code review for request parameters passed into socket, HTTP, or Unix-socket connection APIs
  • Searches for URI parsing without scheme, host, port, address-range, or socket-path allowlists
  • API gateway policies that reject unexpected URI schemes and backend override parameters
  • Network monitoring for unusual outbound connections from the Wyoming process or container
  • Alerts for requests containing tcp://, unix://, loopback, private-range, link-local, or metadata-service destinations
  • Automated security testing that verifies backend selection cannot be redirected through user-controlled input

Sources

SourceLink
github.comgithub.com/OHF-Voice/wyoming/releases/tag/v1.10.2
vulncheck.comvulncheck.com/advisories/wyoming-ssrf-via-uri-query-parameter

Frequently asked

questions

Yes. An attacker generally needs network access to the exposed HTTP API, but the description indicates that authentication is not required.

The attacker must be able to send requests to an affected Wyoming endpoint. They do not necessarily need access to the host or backend configuration.

Do not allow request parameters to select arbitrary connection destinations. If backend selection is necessary, enforce a strict allowlist of approved schemes, hosts, ports, and socket paths.

Treat all client-supplied destinations as untrusted, apply outbound network restrictions, block sensitive address ranges where appropriate, and continuously monitor unexpected egress from application services.

Find out what's already
exploitable in your codebase.