All reports
High

ogx: OGX: Information disclosure via Server-Side Request Forgery in MCP tool server_url

Affected versions

Affected versions
Unknown
Fixed version
Not reported
Affected versions
Unknown
Fixed version
Not reported
Affected versions
Unknown
Fixed version
Not reported

Exploitation

CVE-2026-85666

Probability of exploitation in the next 30 days: 0.73%

FIRST EPSS · 2026-11-20

Description

OGX (formerly Llama Stack, affected at commit fbe8e0f) contains an unauthenticated server-side request forgery vulnerability in the OpenAI-compatible POST /v1/responses endpoint. MCP tool definitions accept a server_url parameter (along with headers and authorization values) that is fetched server-side without destination validation; the existing validate_url_not_private() guard used for other URL inputs is not applied to server_url. On the default starter configuration, which runs without authentication, a remote unauthenticated attacker can cause the server to open connections to arbitrary internal addresses (including cloud metadata endpoints such as http://169.254.169.254/) and forward attacker-supplied headers and bearer tokens to those destinations.

A flaw was found in OGX, an OpenAI-compatible server. A remote, unauthenticated attacker can exploit a Server-Side Request Forgery (SSRF) vulnerability by manipulating the server_url parameter within MCP tool definitions. This allows the attacker to force the server to connect to arbitrary internal network addresses, potentially disclosing sensitive information such as cloud metadata or internal credentials.