Verba CVE-2026-65317

Exploitation requires network reachability to the Verba 2.1.3 /api/connect endpoint and a crafted Origin header prefixed with a single-quote character to defeat the localhost same-origin middleware check; the attacker then supplies arbitrary host and port parameters that the server fetches. Per CVSS PR:N/UI:N no authentication or user interaction is needed, so any exposed instance is reachable. The main limiting factors are network exposure (an instance bound only to localhost or firewalled from attackers is not remotely reachable) and the demonstrated primitive being outbound GET requests, which constrains impact to what a proxied/blind SSRF GET can achieve against reachable internal or external targets.

The provided CVSS 4.0 score is 9.2 (Critical) with vector AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:H/SI:N/SA:N, indicating fully remote, low-complexity, unauthenticated exploitation with high confidentiality impact on both the vulnerable system and a subsequent system (the SSRF target). These signals are internally consistent with the description, which confirms an unauthenticated network attack via a crafted Origin header. However, the description characterizes the demonstrated primitive as outbound GET requests to attacker-controlled infrastructure, which is fundamentally a blind/proxied SSRF; whether an attacker can also read internal service responses (justifying VC:H) depends on how Verba surfaces the response, so the true confidentiality impact should be verified against the vendor advisory. No EPSS score and no KEV status were supplied, so real-world exploitation frequency cannot be quantified; there is no evidence of active exploitation. Given a public technical write-up documenting the bypass, the practical priority is elevated for internet-exposed Verba instances even without confirmed in-the-wild activity.

An attacker who can reach a Verba 2.1.3 instance over the network sends an HTTP request to /api/connect with an Origin header beginning with a single quote to satisfy the flawed localhost check, then sets the host and port parameters to a target of their choosing. The server dutifully issues an outbound GET to that target, letting the attacker probe internal services behind the firewall, hit cloud metadata endpoints, or relay requests through the server's identity. A detailed public write-up documents the bypass technique, lowering the effort required to reproduce it.