CVE-2026-54032

LibreChat MCP descriptions executed script in configuration dialogs

LibreChat-AI · LibreChat

First documented disclosure · Review checkpoint

Summary

Remote MCP metadata could become executable HTML in LibreChat.

Severity and exploitation evidence

Medium · 6.1/10 · CVSS 3.1. Severity source.

Exploitation not established: Public evidence describes a weakness, source correction or controlled demonstration; malicious exploitation against users is not established.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

Unsanitized description HTML reaches dangerouslySetInnerHTML; admin banner HTML has a related sink.

Conditions required

A malicious MCP description and a user opening its configuration dialog; banner path requires administrative control.

Potential impact

Browser script execution in the user’s LibreChat origin.

Affected versions and configurations

Before 0.8.6.

Fix and mitigation status

0.8.6.

Practical next steps

Upgrade to 0.8.6 or later and verify the deployed application image includes the correction.

Operating lesson

Sanitize externally supplied HTML at each rendering sink.

Evidence and limits

Reporter demonstration is a browser-sink reproduction, not a new full-deployment exploit; broad cookie/session theft claims are not independently verified. CNA publication 2026-11-20 is distinct from the earliest reviewed public source. GitHub_M assessment: CVSS 3.1 6.1.

Sources