Stored XSS in mermaid diagrams

Vulnerability Description

HedgeDoc is a collaborative platform for writing and sharing markdown. In HedgeDoc before version 1.7.1 an attacker can inject arbitrary `script` tags in HedgeDoc notes using mermaid diagrams. Our content security policy prevents loading scripts from most locations, but `www.google-analytics.com` is allowed. Using Google Tag Manger it is possible to inject arbitrary JavaScript and execute it on page load. Depending on the configuration of the instance, the attacker may not need authentication to create or edit notes. The problem is patched in HedgeDoc 1.7.1. As a workaround one can disallow `www.google-analytics.com` in the `Content-Security-Policy` header. Note that other ways to leverage the `script` tag injection might exist.

References

• https://github.com/hedgedoc/hedgedoc/releases/tag/1.7.1
• https://github.com/hedgedoc/hedgedoc/security/advisories/GHSA-g6w6-7xf9-m95p
• https://github.com/hackmdio/codimd/issues/1630
• https://github.com/Alemmi/ctf-writeups/blob/main/hxpctf-2020/hackme/solution.md
• https://github.com/hedgedoc/hedgedoc/commit/58276ebbf4504a682454a3686dcaff88bc1069d4