Server-Side Request Forgery in Portkey AI Gateway

Portkey AI Gateway versions 1.15.2 and earlier are affected by a server-side request forgery (SSRF) vulnerability. The flaw exists within the /v1/proxy/* API route, which fails to enforce requestValidator middleware. An attacker can craft malicious HTTP requests that set the 'x-portkey-custom-host' header to point to internal network addresses. When processed, the gateway forwards these requests to the specified internal target. By including Authorization headers in these forged requests, attackers can interact with internal APIs or services not intended for public access. This vulnerability poses a significant risk for the exfiltration of sensitive provider API keys or other internal infrastructure data. Defenders should prioritize patching to version 1.15.3 or later as soon as it becomes available and monitor traffic patterns targeting proxy endpoints.

Impact

Successful exploitation allows unauthenticated attackers to bypass perimeter security to access internal infrastructure. This can result in the exfiltration of sensitive credentials such as provider API keys, unauthorized access to internal management interfaces, and reconnaissance of the local network environment.

Recommendation

• Upgrade Portkey AI Gateway to the latest patched version immediately.
• Implement strict ingress filtering and egress controls on the gateway server to limit connections to internal IP ranges (RFC 1918) and sensitive metadata services (e.g., 169.254.169.254).
• Deploy the provided webserver detection rule to identify anomalous header usage associated with this SSRF vector.