CVE-2026-54031

LibreChat MCP OAuth redirects bypassed destination checks

LibreChat-AI · LibreChat

First documented disclosure · Review checkpoint

Summary

OAuth requests could be redirected to private or internal destinations.

Severity and exploitation evidence

Medium · 6.3/10 · 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

OAuth fetch validated only its initial URL and followed redirects without equivalent validation.

Conditions required

An attacker-controlled configured/shared MCP OAuth endpoint or permission to create one; a user initiates OAuth.

Potential impact

Internal requests and possible response/credential exposure depending on the OAuth step and redirects.

Affected versions and configurations

Before 0.8.8.

Fix and mitigation status

0.8.8.

Practical next steps

Upgrade to 0.8.8 or later and verify the deployed application image includes the correction.

Operating lesson

Apply destination policy consistently to every protocol request and redirect.

Evidence and limits

Default regular users lack MCP_SERVERS.CREATE; administrators or deployments enabling it can create servers. The actual fix rejects redirects and adds literal-IP checking. CNA publication 2026-11-20 is distinct from the earliest reviewed public source. GitHub_M assessment: CVSS 3.1 6.3.

Sources