Openclaw EUVD-2026-17001

The CVSS 6.9 score reflects a local attack vector with low complexity and low privilege requirements (PR:L indicates authenticated access is required), paired with high integrity and availability impact (VI:L, VA:H). The CVSS vector shows no confidentiality impact (VC:N) but significant changes to integrity and availability of the affected resource. While the CVSS score is moderate-to-high, the real-world risk depends on the prevalence of operator.write accounts in typical OpenClaw deployments and the sensitivity of conversation sessions being reset. The local attack vector (AV:L) constrains the threat to users with prior authenticated access to the system, reducing the immediate blast radius compared to network-based exploits. No EPSS score, KEV status, or public proof-of-concept is indicated in the provided data, suggesting this is a newly disclosed vulnerability with limited public validation of exploitability at the time of analysis. Organizations running OpenClaw should treat this as a privilege escalation concern requiring prompt patching, particularly if they deploy OpenClaw in multi-tenant or shared-operator environments where scope separation is a security boundary.

An attacker with legitimate operator.write credentials (e.g., a junior operator or third-party integration user) crafts an agent request containing the /reset slash command targeting a specific conversation session. Due to insufficient authorization checks, the slash command processor executes the reset operation without verifying that the caller holds operator.admin privileges. The attacker successfully clears or reinitializes the conversation state, causing loss of data or service disruption. This scenario requires prior authenticated access to OpenClaw, which limits exploitation to insider threats or compromised operator accounts.