Summary
CVE-2026-53834 is a high-severity authorization bypass vulnerability (CWE-863: Incorrect Authorization) affecting OpenClaw, an open-source AI agent platform, in all versions prior to 2026.4.27. The flaw resides in the QQBot channel plugin’s pre-dispatch slash command handling, where the configured allowFrom access control policy is not applied before a command is dispatched — allowing senders who would otherwise be blocked to trigger slash command execution. The vulnerability carries a CVSS 4.0 score of 8.2 (HIGH) and requires no privileges on the OpenClaw server to exploit.
Technical details
- Root cause: Incorrect authorization (CWE-863) in the QQBot plugin’s pre-dispatch pipeline. When a slash command is received, the
allowFrompolicy check is bypassed before the command is dispatched, meaning restricted senders can invoke commands that should be blocked by the operator’s configured access controls. - Trigger conditions: Exploitation requires that the operator has configured
allowFromaccess control policies and is running QQBot with slash commands enabled. The severity of impact is determined by which slash commands are deployed; as an AI agent framework, OpenClaw can expose capabilities including system command execution, file access, and external API calls. - Attack vector: Fully network-accessible. No privileges on the OpenClaw server are required (PR:N). Any QQ messaging platform user with network access to an exposed OpenClaw instance can send slash commands that reach the pre-dispatch handler before policies are enforced.
- Impact: Unauthorized integrity modification — blocked senders can successfully dispatch slash commands before
allowFromrestrictions are applied. Confidentiality impact is none; integrity impact is high (VI:H per CVSS 4.0).
Affected software
- OpenClaw (
npm: openclaw) — all versions prior to 2026.4.27
Severity
CVSS v3.1 Base Score: 7.5 (HIGH)
Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:N
Mitigation and recommended actions
- Immediate: Upgrade OpenClaw to version 2026.4.27 or later, which contains the vendor-provided fix for this vulnerability.
- If immediate patching is not feasible:
- Restrict network-level access to QQBot slash command endpoints.
- Avoid sharing a single OpenClaw Gateway instance between untrusted users.
- Maintain narrow channel and tool allowlists, and remove any wildcard
allowFromentries where admin-restricted commands are deployed. - Disable QQBot slash commands entirely when they are not actively required.
IONIX Status
The IONIX research team is tracking ongoing exploitation attempts and recommends immediate patching. Potentially affected assets are outlined in this post.

