Openclaw CVE-2026-32899

The CVSS 4.3 score reflects a low-to-moderate severity profile with network attack vector, low complexity, and low privilege requirements, but limited impact (integrity only, no confidentiality or availability impact). The vulnerability requires prior authentication (PR:L), reducing the attack surface compared to unauthenticated exploits. While CVSS alone might suggest lower priority, the authorization bypass nature (CWE-863) combined with low attack complexity and consistency with multi-vector intelligence (VulnCheck, GitHub advisory, multiple commits indicating deliberate fixes) indicates this represents a real security gap in access control enforcement. The availability of working patches (two commits referenced) suggests the vendor confirmed the issue and implemented fixes, elevating credibility. The exploitation probability is moderate due to the requirement for authenticated access, but the automatable nature of event injection and bypass of intended policy controls makes this a genuine priority for organizations using OpenClaw in security-sensitive contexts, particularly where DM policies enforce sensitive communication restrictions.

An authenticated attacker in an OpenClaw workspace can craft reaction_* or pin_* events spoofing a restricted sender (one blocked by configured DM policies or channel allowlists) and inject these events into the system-event context, bypassing the intended access controls. Since the attack requires only network access and low privileges (existing user account), an insider or compromised low-privilege account can bypass channel member restrictions and DM policies in under a minute by sending malformed or manipulated event payloads. The attacker gains no direct data access or system disruption, but achieves unauthorized event injection and permission-model circumvention, enabling them to perform actions they should not be authorized for.