Executive brief
Aix-DB is a data analytics platform that uses markdown to render chat responses, skill descriptions, and knowledge messages. The application fails to sanitize HTML content in markdown before displaying it in user interfaces, allowing attackers to inject malicious scripts that execute in victims' browsers. An attacker with the ability to inject content (such as a compromised skill or message) can steal user sessions, redirect users, or perform actions on their behalf.
Technical details
The vulnerability is a stored cross-site scripting (XSS) flaw in Aix-DB's markdown rendering component. The markdown parser is configured with `html: true`, enabling raw HTML passthrough, and this HTML is bound directly to Vue.js v-html directives without sanitization. The vulnerable code is located in the MarkdownPreview component's markdown.ts plugin, which creates a MarkdownIt instance that renders markdown content as raw HTML. Attackers can inject malicious JavaScript through markdown content in chat responses, skill descriptions, or knowledge messages. When users view this content, the malicious scripts execute in their browser context. No authentication bypass or special privilege is required to inject content in certain contexts. A fix requires either disabling raw HTML in the markdown parser or implementing proper HTML sanitization before rendering.
Affected products
- Aix-DB Aix-DB through 1.2.4
Timeline
- 2026-08-31: disclosed
- 2026-08-31: advisory: CVE-2026-82881 published