A vulnerability was identified in SciPhi-AI R2R up to 3.6.6. (CVE-2026-105148)

CVE-2026-105148 describes a server-side request forgery vulnerability in SciPhi-AI R2R versions up to 3.6.6. The vulnerability is located in the Retrieval Completion API Endpoint component, specifically in the file py/shared/abstractions/llm.py. Manipulating the generation_config.api_base argument allows an attacker to induce the server to make unauthorized requests. The vulnerability can be exploited remotely without privileges or user interaction. Although exploit code is publicly available, the vendor has not responded to disclosure and no patch or remediation guidance is currently available.

Successful exploitation of this vulnerability allows an unauthenticated remote attacker to perform server-side request forgery, potentially enabling unauthorized internal network access, data exfiltration, or other impacts depending on the server's network environment. The CVSS v3.1 score is 7.3 (medium severity) reflecting low complexity, no privileges required, and impacts to confidentiality, integrity, and availability.

Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. Since the vendor has not responded and no official fix is available, users should consider implementing network-level protections to restrict outbound requests from the affected service and monitor for suspicious activity related to the vulnerable API endpoint.