Openclaw EUVD-2026-17007

CVSS 6.9 with CVSS:4.0 vector (AV:N, AC:L, PR:N, UI:N) indicates network-accessible, unauthenticated, low-complexity exploitation with partial confidentiality and integrity impact (VC:L, VI:L). EPSS data is not provided, but the straightforward nature of the bypass (omitting a single field in an inbound event) suggests moderate-to-high exploitation probability. No CISA KEV status is indicated, so active exploitation is not confirmed; however, the attack does not require sophisticated tooling-a malicious Feishu user or compromised bot integration can trivially craft reaction events without chat_type. The impact is bounded to unauthorized access within group chats where reaction-based controls are relied upon; it does not yield remote code execution or full system compromise. Organizations using OpenClaw with strict group chat access policies are at genuine risk, but the exposure is limited to the subset of deployments enforcing those controls. This is a real priority for affected deployments, not merely a theoretical high-CVSS issue.

A malicious Feishu user, or an attacker with access to a compromised Feishu workspace, crafts a reaction event (e.g., emoji reaction to a message) and deliberately omits the chat_type field in the event payload. OpenClaw misinterprets this as a p2p event and bypasses group-level access controls (groupAllowFrom, requireMention). The attacker can then trigger bot actions or access information that should be restricted to authorized group members, without meeting the group's explicit access policy requirements. No public proof-of-concept code is known to exist, but the attack is achievable through direct Feishu API manipulation or by controlling a bot integration that sends crafted events.