Executive brief
OpenClaw, a tool for managing automated tasks and skill commands, contains a flaw where certain commands can bypass security and auditing checks. This means that specific actions could be performed without being logged or verified against established safety policies. While the risk is primarily internal to authenticated users, it could allow an operator to circumvent corporate oversight or auditing requirements.
Technical details
A protection mechanism failure (CWE-693) exists in OpenClaw's skill-command dispatch logic. Specifically, certain dispatch paths fail to trigger the 'runBeforeToolCallHook', which is responsible for auditing and policy enforcement before a tool is executed. An authenticated attacker with network access can exploit this to run commands that bypass hook-based security controls. The vulnerability is addressed in version 2026.5.6; until then, users should avoid relying solely on hooks for enforcement and should narrow tool allowlists.
Affected products
- OpenClaw openclaw <= 2026.5.5
Timeline
- 2026-05-28: disclosed: Initial disclosure by researcher
- 2026-06-18: advisory: GitHub Advisory published
- 2026-05-28: patched: Version 2026.5.6 released