OpenClaw Improper Trust Boundary Vulnerability Leading to Unintended Code Execution

A vulnerability exists in OpenClaw versions prior to 2026.4.2, where untrusted workspace channel shadows can execute during the built-in channel setup and login processes. This flaw allows attackers to clone a workspace with a malicious plugin that claims a bundled channel ID, enabling unauthorized in-process code execution before the plugin is explicitly trusted. The issue arises from an improper trust boundary that allows unverified workspace channel metadata to override the default channel setup and login flows.

Impact

Exploitation of this vulnerability could result in unauthorized code execution within the application's process, bypassing the intended trust model for workspace plugins.

Reproduction

To reproduce this vulnerability, create a workspace plugin that claims a bundled channel ID but is not trusted. During the channel setup process, the untrusted plugin can execute code, taking advantage of the trust boundary flaw. This can be tested by cloning a workspace and introducing a malicious plugin that exploits the vulnerability.

Remediation

Users can update to OpenClaw version 2026.4.2 or later to address this vulnerability.