OpenClaw Feishu bug ignores disabled permissions

Published September 3, 2026 GHSA-w8wf-3qvj-6xqf

OpenClaw is an open-source AI assistant that connects to chat apps and can act on a user's behalf. A bug in its Feishu integration let permission tools run actions even after an operator had turned them off for a specific account, so a disabled account could still trigger things it should not have been able.

Severity
HighCVSS 8.1
Fix
Fixed in 2026.6.9Fix recorded on Sep 3, 2026
Affected versions
before 2026.6.9-beta.1
Weakness
CWE-863
Affects
@openclaw/feishu

How it works

The Feishu permission tool checked whether a feature was allowed in general but did not properly check the per-account disablement setting, so a lower-trust caller or a configured input path could reach actions that should have needed stronger authorization.

What to do

Check the installed version of the @openclaw/feishu package on your OpenClaw Gateway and compare it against 2026.6.9-beta.1. If it predates that release and the Feishu permission tools are enabled and reachable, you are affected.

Run this in the application environment you want to check:

npm ls @openclaw/feishu

Upgrade to OpenClaw 2026.6.9 or later, and until you can update, restrict the affected Feishu permission feature to trusted operators or disable it, keep channel and tool allowlists narrow, and avoid sharing one Gateway between mutually untrusted users.

Technical details

Affected software: @openclaw/feishu

OpenClaw Feishu permission tools could ignore per-account disablement. Feishu permission tools could ignore per-account disablement.

References

View Original Source