On this page (7 sections)
A critical stored cross-site scripting (XSS) flaw, tracked as CVE-2026-130681, affects SiYuan before version 3.8.10. According to the GitHub Advisory Database entry, the bug can lead to operating system commands running through Node.js when a desktop user opens a particular panel. It is described as an incomplete fix for an earlier issue, CVE-2026-44670.
If you use the SiYuan desktop app, check your version now and update to 3.8.10 or later. The advisory is short and several details are still unknown, so this article separates what the advisory says from general guidance.
What happened
The advisory was published on November 18, 2026. It says SiYuan before 3.8.10 has a stored XSS issue because a function named renderAVAttribute() places an Attribute View name into the header of the database row attribute panel without escaping it.
Because the name is stored and later rendered as HTML, an attacker who can persist a crafted name can inject an onerror handler that runs when the panel is opened. The advisory gives the setAttrViewName action as one example of how the name can be persisted, and says the handler can run OS commands through Node.js when a desktop user opens the panel.
The flaw is described as an incomplete fix for CVE-2026-44670, which suggests an earlier patch did not cover this rendering location. The research for this story does not describe what CVE-2026-44670 itself covers, so treat that link as context only.
Why a stored XSS can become command execution
In an ordinary website, XSS lets an attacker run script in a victim's browser tab, which is serious but limited by the browser sandbox. Here the advisory says the injected handler can reach Node.js on the desktop app, which is why a text-rendering mistake carries a 9.4 score.
The underlying weakness class is simple: untrusted text is inserted into a page as markup instead of plain text. The standard defense is to escape or sanitize every user-controlled value where it is rendered, not only where it is saved.
The CVSS v4 metrics in the advisory list network attack vector, low attack complexity, no attack requirements, low privileges required, passive user interaction, and high impact on confidentiality, integrity and availability for both the vulnerable and subsequent systems. The advisory does not spell out deployment scenarios beyond this.
Who is realistically exposed
The advisory itself only states that an attacker who can persist an Attribute View name can affect a desktop user who opens the panel. The points below are our general interpretation, not statements from the advisory:
- Desktop users on versions before 3.8.10 are the group the advisory describes.
- Exposure is likely higher where content such as database names can come from someone other than you, for example data created by another person. The advisory does not say which sharing or import features make this possible.
- A user who only works with their own content has a lower practical risk, but updating is still sensible.
How to check your exposure
Find the version you run. In the desktop app, open the About or settings area and read the version number. If it is lower than 3.8.10, you are in the affected range described by the advisory.
Also consider where your SiYuan content comes from. If other people can write to your data, or you have brought in content from elsewhere, treat Attribute View names as untrusted until you have updated.
What to do now
- Find every SiYuan desktop installation you or your team use and note its version.
- Update each one to version 3.8.10 or later, following the SiYuan project's release notes and upgrade instructions.
- Until you have updated, avoid opening database row attribute panels for content that someone else could have edited.
- Review who can write to your SiYuan data and remove access that is no longer needed.
- Look through Attribute View names for anything unusual, such as angle brackets or HTML-like text, and restore from a known-good backup if you find tampering.
- If you suspect a desktop was exposed, treat it as a possible compromise: rotate passwords, tokens and keys stored on that machine, and check for unexpected processes.
Detection and monitoring afterwards
A stored XSS leaves its payload in your data, so updating the app does not necessarily remove content that was already planted. After patching, review Attribute View names and other titles for unexpected HTML, and compare against a backup from before the disclosure if you have one.
On desktop machines, watch for signs of command execution: unexpected child processes started by the SiYuan app, new scheduled tasks or startup entries, and outbound connections you do not recognize. Endpoint security tooling is the right place for this.
If you also run web-facing services, general hygiene helps: keep certificates healthy with the free SSL checker, and read what a WAF does to understand where it helps. Webforta Monitor can watch certificate and header changes. None of these replace patching the application itself.
What we still do not know
- Whether the flaw is being exploited. The sources do not say either way.
- The exact first fixed release beyond the phrase "before 3.8.10", since the advisory lists patched versions as unknown.
- Whether other ways of running SiYuan, such as a browser-based deployment, are affected. The advisory only describes command execution for desktop users.
- Technical detail on CVE-2026-44670, the earlier issue this one follows up.
We will not speculate beyond the advisory. If the SiYuan project publishes release notes or more detail, rely on those over this summary.
Sources
This article is based on the public sources below. Details can change as vendors and researchers publish more; check the original advisories before acting on version numbers.
