Executive brief
A security vulnerability exists in LoLLMs, an open-source platform for Large Language Models. The software fails to properly clean user-provided message content, allowing an attacker to inject malicious scripts that execute in other users' browsers. This could lead to unauthorized access to user accounts, session hijacking, or the spread of malicious code within the platform.
Technical details
A stored Cross-site Scripting (XSS) vulnerability exists in the `from_dict` method of the `AppLollmsMessage` class in parisneo/lollms prior to version 2.2.0. The root cause is a failure to sanitize or HTML-encode the `content` field during the deserialization of user-provided data. An unauthenticated remote attacker can exploit this by providing a crafted dictionary containing malicious HTML or JavaScript payloads. When this content is rendered in a victim's browser, the script executes in the context of the application's origin, potentially leading to session hijacking or account takeover. The issue was addressed in version 2.2.0 by integrating the `bleach` library for input sanitization.
Affected products
- parisneo lollms < 2.2.0
Timeline
- 2026-01-17: patched: Fix committed to repository
- 2026-04-12: disclosed: CVE published