Executive brief
A security vulnerability exists in the social networking feature of LoLLMs, a platform for large language models. Attackers can post malicious content that, when viewed by other users or administrators, executes hidden scripts in their web browsers. This could allow an attacker to take over user accounts, steal login sessions, or spread malicious activity across the platform.
Technical details
A stored Cross-Site Scripting (XSS) vulnerability exists in the `create_post` function within `backend/routers/social/__init__.py` of the LoLLMs platform. The application fails to sanitize user-provided content before assigning it to the `DBPost` model and subsequently rendering it in the Home Feed. A remote, unauthenticated attacker can inject malicious JavaScript that executes in the context of any user viewing the feed, including administrators. This can lead to session hijacking, account takeover, and wormable attacks. The vulnerability is mitigated in version 2.2.0 by implementing input sanitization using the `bleach` library.
Affected products
- parisneo lollms < 2.2.0
Timeline
- 2026-01-17: patched: Sanitization implemented in commit 9767b882dbc893c388a286856beeaead69b8292a
- 2026-04-10: disclosed
- 2026-04-10: advisory