Executive brief
A security vulnerability exists in LoLLMs, an open-source platform for large language models, specifically within its direct messaging and prompt-sharing features. An attacker can send a specially crafted message that, when opened by another user, executes malicious code in their web browser. This could allow the attacker to steal sensitive data, perform actions as the victim, or take over the victim's account.
Technical details
A stored cross-site scripting (XSS) vulnerability exists in the `POST /api/prompts/share` endpoint of parisneo/lollms. The application fails to perform server-side sanitization of the `prompt_content` field before storing it in the database. On the client side, the `MessageContentRenderer` component uses Vue's `v-html` directive to render these messages, relying on a regex-based sanitizer that is insufficient to block malicious HTML payloads. An authenticated attacker can exploit this by sending a malicious prompt-share message to a victim; when the victim views the message, the payload executes. This can lead to session hijacking, unauthorized API calls on behalf of the victim, and full account takeover.
Affected products
- parisneo lollms latest
Timeline
- 2026-07-18: disclosed
- 2026-07-18: advisory