CVE-2026-53835: Openclaw Auth Bypass Vulnerability

CVE-2026-53835 Overview

CVE-2026-53835 is an authorization bypass vulnerability in OpenClaw versions prior to 2026.5.6. The flaw resides in the Feishu dynamic-agent binding feature, which fails to enforce configured config-write controls. Authenticated senders can create or update sender-agent bindings without honoring the intended policy boundaries. The weakness is classified as [CWE-863] Incorrect Authorization. Exploitation requires valid authentication and network access to the OpenClaw service. The vulnerability allows unauthorized modification of binding state but does not directly expose data confidentiality or system availability.

Critical Impact
Authenticated attackers can bypass config-write enforcement in OpenClaw Feishu integrations to modify dynamic-agent bindings beyond their authorized policy scope.

Affected Products

OpenClaw (Node.js distribution) versions prior to 2026.5.6
Deployments using the Feishu dynamic-agent binding feature
OpenClaw instances exposing sender-agent binding APIs to authenticated users

Discovery Timeline

2026-06-12 - CVE-2026-53835 published to NVD
2026-06-17 - Last updated in NVD database

Technical Details for CVE-2026-53835

Vulnerability Analysis

The vulnerability stems from missing authorization enforcement in the Feishu dynamic-agent binding code path within OpenClaw. OpenClaw supports configuration-driven write controls that determine which authenticated principals may alter binding state. The Feishu dynamic-agent binding handler does not consult these config-write controls before applying changes. As a result, an authenticated sender can create or update bindings that the configured policy was meant to forbid. The defect maps to [CWE-863] Incorrect Authorization, where access decisions are made on incomplete or incorrect criteria.

Root Cause

The root cause is a missing authorization check between request acceptance and binding mutation. OpenClaw's dynamic-agent binding logic treats authentication as sufficient evidence of authorization. The configured config-write policy is not evaluated against the requesting sender before the binding state changes. This logic gap allows any authenticated sender to drive the binding feature beyond its intended administrative scope.

Attack Vector

An attacker with valid OpenClaw credentials sends a network request to the Feishu dynamic-agent binding endpoint. The request specifies a sender-agent binding creation or update that would normally be blocked by config-write enforcement. Because the handler skips the policy check, the binding is created or modified. The impact is limited to integrity of binding configuration; the vulnerability does not yield code execution or data disclosure on its own. Refer to the VulnCheck Advisory on Feishu Bypass and the GitHub Security Advisory for technical details.

Detection Methods for CVE-2026-53835

Indicators of Compromise

Unexpected creation or update events for Feishu sender-agent bindings outside scheduled administrative windows.
Binding modifications attributed to authenticated senders that lack documented config-write privileges.
Repeated API calls to the dynamic-agent binding endpoint from a single sender identity within a short interval.

Detection Strategies

Audit OpenClaw application logs for binding create or update operations and correlate them against the configured config-write policy.
Compare the current binding state to a known-good baseline and alert on divergence.
Track which authenticated principals invoke the Feishu dynamic-agent binding API and flag principals not on an allowlist.

Monitoring Recommendations

Forward OpenClaw audit logs to a centralized log platform with retention sufficient for incident review.
Generate alerts on binding state changes that occur without a corresponding administrative ticket or change request.
Review authentication logs alongside binding-change events to detect compromised sender credentials.

How to Mitigate CVE-2026-53835

Immediate Actions Required

Upgrade OpenClaw to version 2026.5.6 or later, which contains the fix for the Feishu dynamic-agent binding bypass.
Inventory current Feishu sender-agent bindings and revert any unauthorized changes to a verified baseline.
Rotate credentials for any sender identities that may have abused the dynamic-agent binding endpoint prior to patching.

Patch Information

The maintainers have released a fixed version. Upgrade to OpenClaw 2026.5.6 or later. Patch details are documented in the GitHub Security Advisory GHSA-3wqp-prf6-2m72.

Workarounds

Disable the Feishu dynamic-agent binding feature until the upgrade is applied, if operationally feasible.
Restrict network access to the OpenClaw binding API to a trusted administrative network segment.
Reduce the set of authenticated senders permitted to reach the dynamic-agent binding endpoint via upstream access controls.

bash

# Configuration example: upgrade OpenClaw via npm to the patched release
npm install openclaw@2026.5.6