Executive brief
OpenClaw is an AI assistant platform. A vulnerability in its channel extensions allows the system to make unauthorized network requests to internal or restricted destinations. This could allow an attacker to bypass network security controls and access sensitive internal services or data that should not be reachable from the internet.
Technical details
Multiple channel extensions in OpenClaw were found to use the raw fetch() API against configured base URLs without proper validation. This represents an incomplete fix for CVE-2026-28476, as it allows outbound requests to be redirected or rebound to blocked internal destinations. An attacker with the ability to configure or influence these base URLs can perform Server-Side Request Forgery (SSRF) to probe or interact with internal network resources. The vulnerability is addressed by routing outbound requests through a new fetchWithSsrFGuard function. The fix is available in version 2026.3.25 and later.
Affected products
- openclaw openclaw <= 2026.3.24
Timeline
- 2026-03-26: disclosed
- 2026-03-29: advisory: GitHub Advisory published