GoClaw EUVD-2026-45383

Exploitation requires two hard prerequisites: the attacker must already hold high-level privileges within GoClaw (PR:H per the CVSS 4.0 vector), and must operate from an adjacent network position - the same LAN, VLAN, or VPN segment where the GoClaw service is reachable (AV:A). The vulnerability is specific to deployments where GoClaw's binary execution approval feature (internal/tools/exec_approval.go) is active and in use. These two conditions together severely narrow the realistic attacker pool to insiders, compromised privileged accounts, or post-compromise lateral movement scenarios. Exploitation is not possible from an arbitrary internet-facing position, and unauthenticated or low-privileged users cannot reach the vulnerable code path. No special non-default configuration beyond having the execution approval feature enabled appears required.

The CVSS 4.0 score of 4.8 (Medium) accurately reflects the constrained threat model. Three signals converge to limit real-world risk: AV:A restricts exploitation to the same network segment (LAN, VLAN, or VPN), PR:H means the attacker must already hold high privileges within GoClaw before the flaw is reachable, and the impact triad VC:L/VI:L/VA:L with SC:N/SI:N/SA:N confirms limited blast radius. The E:P modifier acknowledges a working PoC, which does elevate urgency for environments where insiders or already-compromised privileged accounts are a realistic threat. The Authentication Bypass tag is accurate but potentially misleading at face value - this is a privilege escalation within an already-privileged session rather than a complete authentication bypass from an anonymous state. No KEV listing exists, and EPSS data is not provided, but the constrained prerequisites suggest targeted rather than widespread exploitation. This is a realistic insider-threat or post-compromise lateral-movement scenario, not a mass-exploitation risk.

An insider or attacker who has already obtained high-privileged GoClaw credentials and is positioned on the same network segment sends a crafted RequestApproval payload targeting the matchesAllowlist function. The flawed authorization logic fails to correctly reject the request, resulting in approval of a binary execution that the allowlist was intended to block. A public proof-of-concept demonstrating this technique is available at https://github.com/nextlevelbuilder/goclaw/issues/1216, meaning the exact manipulation required is publicly documented.