Executive brief
LibreChat RAG API, a service used to process and search through uploaded documents using AI, contains a flaw that allows users to inject fake entries into system logs. This could allow a malicious actor to hide their activities, frame other users, or disrupt digital forensics during an investigation. If these logs are viewed through certain management tools, it could also lead to further technical attacks on administrative systems.
Technical details
A log-injection vulnerability exists in the LibreChat RAG API (version 0.7.0) due to improper sanitization of user-supplied input before it is written to system logs. Specifically, an attacker can insert Carriage Return and Line Feed (CRLF) characters into the 'file_id' parameter of a POST request to the FastAPI-based backend. This allows for the creation of forged log entries, which can be used to obfuscate malicious activity or misdirect forensic investigations. While the CVSS vector provided by CISA-ADP suggests no authentication is required (PR:N), the advisory text mentions 'authenticated attackers' in the impact description. As of the disclosure date, no official patch is available; mitigations include implementing input sanitization filters and validating RAG output before it is processed by downstream log-management tools.
Affected products
- LibreChat LibreChat RAG API 0.7.0
Timeline
- 2026-01-26: other: Vendor notified
- 2026-03-16: disclosed: Public disclosure by CERT/CC
- 2026-03-16: advisory: NVD publication date