Skip to content

Improper sanitization of architecture diagram iconText leads to XSS

Critical
sidharthv96 published GHSA-8gwm-58g9-j8pw Aug 19, 2025

Package

npm mermaid (npm)

Affected versions

>=11.1.0, <=11.9.0

Patched versions

11.10.0

Description

Summary

In the default configuration of mermaid 11.9.0, user supplied input for architecture diagram icons is passed to the d3 html() method, creating a sink for cross site scripting.

Details

Architecture diagram service iconText values are passed to the d3 html() method, allowing malicious users to inject arbitrary HTML and cause XSS when mermaid-js is used in it's default configuration.

The vulnerability lies here:

export const drawServices = async function (
  db: ArchitectureDB,
  elem: D3Element,
  services: ArchitectureService[]
): Promise<number> {
  for (const service of services) {
    /** ... **/
    } else if (service.iconText) {
      bkgElem.html(
        `<g>${await getIconSVG('blank', { height: iconSize, width: iconSize, fallbackPrefix: architectureIcons.prefix })}</g>`
      );
      const textElemContainer = bkgElem.append('g');
      const fo = textElemContainer
        .append('foreignObject')
        .attr('width', iconSize)
        .attr('height', iconSize);
      const divElem = fo
        .append('div')
        .attr('class', 'node-icon-text')
        .attr('style', `height: ${iconSize}px;`)
        .append('div')
        .html(service.iconText); // <- iconText passed into innerHTML
       /** ... **/
};
};

This issue was introduced with 734bde3, around 15 months ago, which was released in v11.1.0.

PoC

Render the following diagram and observe the modified DOM.

architecture-beta
    group api(cloud)[API]
    service db "<img src=x onerror=\"document.write(`xss on ${document.domain}`)\">" [Database] in api

Here is a PoC on mermaid.live: https://mermaid.live/edit#pako:eNo9T8FOwzAM_ZXI4rBJpWrpRtuIISF24caZZdKyxOsiLUnlJjCo-u9kQ8wX-_n5-dkjKK8ROEhSRxNQhUh4v8cghWMpOvKxZ7I3M3XyUc83L-9v2z9qQPo0CpneMwFPxnZsILU6M--QyNNKCAHaq2jRhfyL0vLZ7jwMiWd3443Q3krjpt38Mv4sgG3WMsi9HHDLjLs4CwcZdGQ08EARM7BISZMgjJdLBIQjWhTAU6nxIOMpCBBuSrJeug_v7b8yPdMdgR_kaUgo9loGXBvZkbS3LqHTSK8-ugC8LMrrEuAjnIEvlnlVL9q6rZu6Lh-rRQbfwKuyyZuybcvqIaWiqKcMfq6uRd7Uy-kXhYFzcA

Impact

XSS on all sites that use mermaid and render user supplied diagrams without further sanitization.

Remediation

Sanitize the value of iconText before passing it to html().

Severity

Critical

CVE ID

CVE-2025-54880

Weaknesses

Improper Neutralization of Input During Web Page Generation ('Cross-site Scripting')

The product does not neutralize or incorrectly neutralizes user-controllable input before it is placed in output that is used as a web page that is served to other users. Learn more on MITRE.

Credits