NewApp Monitoring is liveSign up free
AgentbeamBeta
AI agent vulnerability · OpenClaw

CVE-2026-35647: OpenClaw: Matrix Verification Notices Bypass Matrix DM Policy and Reply to Unpaired DM Peers

mediumCVSS 4.0 5.3GHSA-9wqx-g2cw-vc7r

OpenClaw 2026.3.24 and earlier lets the Matrix extension send verification notices to direct-message peers who are unpaired or blocked by the DM policy. A Matrix user the agent should ignore can still make the bot reply. The advisory names 2026.3.25 as the first patched version.

Published Mar 27, 2026 · Updated Apr 10, 2026 · Analysis Oct 31, 2026

What an attacker can do

OpenClaw's Matrix channel applies a DM policy (dmPolicy, allowFrom, and the pairing store) to decide who may talk to the agent. The verification event router in extensions/matrix/src/matrix/monitor/verification-events.ts posted notices for incoming verification requests and SAS summaries without consulting that policy. Verification is an alternate path around the DM check (CWE-288).

An attacker needs a Matrix account that can start a DM or a verification request with the bot, plus network access to the homeserver. They get replies from the bot that the policy was meant to suppress. This confirms the bot is live, and it lets the attacker generate bot messages in a DM room. The CVSS 4.0 vector shows low integrity impact only, with no confidentiality or availability effect.

The advisory does not describe any command execution or data exposure. The risk is mostly for teams that run an OpenClaw Matrix bot with a restrictive dmPolicy and rely on it to keep strangers out.

Who is affected

The npm package openclaw at 2026.3.24 and earlier, with the Matrix extension enabled. Deployments that use dmPolicy values other than open rely on the check that was bypassed. The advisory verified the flaw on tag v2026.3.24.

Check your version:

npm ls openclaw

openclaw

Ecosystem
npm
Affected
<= 2026.3.24
Fixed in
No patched version yet
Check
npm ls openclaw

How to fix it

Upgrade to openclaw 2026.3.25 or later. The package metadata here lists no patched version, but the advisory names 2026.3.25, so confirm it is published on npm.

npm install openclaw@latest

The fix is commit 2383daf. Verification notices are now blocked when DMs are disabled or dmPolicy is disabled. They are allowed when it is open. In other modes the sender must match allowFrom or the pairing store. Blocked notices are written to the verbose log. The advisory gives no workaround.

How to tell if you were hit

Look for Matrix DM rooms where the bot posted verification or SAS emoji notices to users who are not in your allowFrom list or the pairing store. On a patched version, blocked notices appear as verbose log entries. Review recent verification requests against the list of paired users.

Original advisory

From the GitHub Advisory Database (CC BY 4.0).

Summary

Matrix Verification Notices Bypass Matrix DM Policy and Reply to Unpaired DM Peers

Affected Packages / Versions

  • Package: openclaw
  • Affected versions: <= 2026.3.24
  • First patched version: 2026.3.25
  • Latest published npm version at verification time: 2026.3.24

Details

Matrix verification notices previously bypassed DM access checks and could reply to peers that were unpaired or otherwise outside the allowed DM policy. Commit 2383daf5c4a4e08d9553e0e949552ad755ef9ec2 gates verification notices on DM access before sending.

Verified vulnerable on tag v2026.3.24 and fixed on main by commit 2383daf5c4a4e08d9553e0e949552ad755ef9ec2.

Fix Commit(s)

  • 2383daf5c4a4e08d9553e0e949552ad755ef9ec2

Credited to zpbrent.

Questions

Can an unpaired Matrix user run commands through this bug?

The advisory does not say so. It describes only that the bot sends verification notices to peers outside the DM policy. The impact rating is low integrity.

What does the fix check?

A new function, isVerificationNoticeAuthorized, runs before a notice is posted. It checks the DM settings and the pairing allow store.

Which version fixes CVE-2026-35647?

2026.3.25, per the advisory. 2026.3.24 and earlier are affected.

References

How this page is made: the data panel comes from the GitHub Advisory Database, CISA’s KEV catalog, the NVD and FIRST EPSS, and updates twice a day. The analysis is written by an AI model from the sources listed above and checked by automated rules before it is published. Check versions against the vendor's advisory before you act. Methodology

More OpenClaw vulnerabilities

All OpenClaw entries