CVE-2026-105148: Server-Side Request Forgery in SciPhi-AI R2R

This vulnerability in SciPhi-AI R2R (up to version 3.6.6) allows an attacker to perform server-side request forgery by manipulating the generation_config.api_base argument in the Retrieval Completion API Endpoint component. The vulnerable code is located in py/shared/abstractions/llm.py. The attack can be launched remotely without authentication. Public exploit code exists, increasing the risk of exploitation. The vendor has not provided any response or patch, and no official remediation is documented.

Successful exploitation of this SSRF vulnerability could allow an attacker to make unauthorized requests from the vulnerable server to internal or external systems. This could lead to information disclosure or interaction with internal services not normally accessible externally. The CVSS 4.0 base score is 6.9 (medium severity), reflecting network attack vector, low complexity, no privileges or user interaction required, but limited confidentiality, integrity, and availability impacts.

No official patch or remediation is currently available from the vendor. Since the vendor has not responded and no fixes are published, users should consider implementing network-level protections such as restricting outbound requests from the affected component to trusted destinations only, if feasible. Monitor for unusual outbound traffic patterns. Check the vendor advisory regularly for updates or patches.