Cross-Site Scripting (XSS) via SVG Schema innerHTML Injection in @pdfme/schemas

Description

Summary

The SVG schema plugin in @pdfme/schemas renders user-supplied SVG content using container.innerHTML = value without any sanitization, enabling arbitrary JavaScript execution in the user's browser.

Details

In packages/schemas/src/graphics/svg.ts, line 87, the SVG schema's ui renderer assigns raw SVG markup directly to innerHTML when in viewer mode or form mode with readOnly: true:

// svg.ts, line 81-94 (non-editable rendering path)
} else {
  if (!value) return;
  if (!isValidSVG(value)) {
    rootElement.appendChild(createErrorElm());
    return;
  }
  container.innerHTML = value;  // <-- VULNERABLE: unsanitized SVG injected into DOM
  const svgElement = container.childNodes[0];
  if (svgElement instanceof SVGElement) {
    svgElement.setAttribute('width', '100%');
    svgElement.setAttribute('height', '100%');
    rootElement.appendChild(container);
  }
}

The isValidSVG() function (lines 11-37) only validates that the string contains <svg and </svg> tags and passes DOMParser well-formedness checks. It does NOT strip or block:
- <script> tags embedded in SVG
- Event handler attributes (onload, onerror, onclick, etc.)
- <foreignObject> elements containing HTML with event handlers
- <animate> / <set> elements with onbegin / onend handlers
- SVG <use> elements referencing malicious external resources

All of these are valid SVG and pass isValidSVG(), but execute JavaScript when inserted via innerHTML.

Attack Vectors

1. Malicious Template (readOnly SVG schema)

An attacker crafts a template JSON with a readOnly SVG schema containing a malicious content value. When loaded into the pdfme Form or Viewer component, the SVG executes JavaScript.

2. Application-Supplied Inputs + Viewer

If an application uses the pdfme Viewer component and passes user-controlled data as inputs for a non-readOnly SVG schema, the attacker's SVG flows directly to innerHTML.

Proof of Concept

Loading the following template into a pdfme Form or Viewer component triggers JavaScript execution:

{
  "basePdf": { "width": 210, "height": 297, "padding": [20, 20, 20, 20] },
  "schemas": [[
    {
      "name": "malicious_svg",
      "type": "svg",
      "content": "<svg xmlns='http://www.w3.org/2000/svg' onload='alert(document.domain)'><rect width='100' height='100' fill='red'/></svg>",
      "readOnly": true,
      "position": { "x": 20, "y": 20 },
      "width": 80,
      "height": 40
    }
  ]]
}

Additional payloads that bypass isValidSVG() and execute JavaScript:

<!-- Via foreignObject -->
<svg xmlns="http://www.w3.org/2000/svg"><foreignObject width="200" height="60"><body xmlns="http://www.w3.org/1999/xhtml"><img src="x" onerror="alert(1)"/></body></foreignObject></svg>

<!-- Via animate onbegin -->
<svg xmlns="http://www.w3.org/2000/svg"><rect width="100" height="100"><animate attributeName="x" values="0" dur="0.001s" onbegin="alert(1)"/></rect></svg>

Impact

An attacker who can supply a malicious template (via file upload, shared template URL, multi-tenant template storage, or updateTemplate() API) can execute arbitrary JavaScript in the context of any user who views or fills the template. This enables:
- Session hijacking via cookie/token theft
- Keylogging of form inputs (including sensitive data being entered into PDF forms)
- Phishing attacks by modifying the rendered page
- Data exfiltration from the application

The attack is particularly concerning for multi-tenant SaaS applications using pdfme where templates may be user-supplied.

Suggested Fix

Sanitize SVG content before DOM insertion using DOMPurify or a similar library:

import DOMPurify from 'dompurify';

// Replace line 87:
container.innerHTML = DOMPurify.sanitize(value, { USE_PROFILES: { svg: true } });

Alternatively, parse the SVG via DOMParser, strip all script elements and event handler attributes, then append the sanitized DOM nodes.

Basic information

Type
reviewed
Severity
medium
Advisory on GitHub
Open advisory ↗
Repository advisory
Open repository advisory ↗
Source code
Browse source ↗
Published (advisory)
2026-03-18 16:10:26 UTC
Updated
2026-03-18 16:10:30 UTC
GitHub reviewed
2026-03-18 16:10:26 UTC

CVSS Scores

Base score Version Severity Vector
6.1 3.1
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:C/C:L/I:L/A:N Click to expand
Attack vector (AV:N)
Could be attacked over the internet or any normal routed network—not just someone sitting at the machine.
Attack complexity (AC:L)
Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup.
Privileges required (PR:N)
No account or special rights needed—anonymous or random user is enough.
User interaction (UI:R)
A real person has to do something—click, install, enable—otherwise it doesn’t land.
Scope (S:C)
Breaking this can reach past the original component and bite other resources—bigger blast radius.
Confidentiality (C:L)
Some sensitive info could get out, but not a total data dump.
Integrity (I:L)
Attackers could change some data, but it’s limited—not everything goes.
Availability (A:N)
Service keeps running; no real outage angle.

Identifiers

Type Value
GHSA GHSA-87v3-4cfp-cm76 ↗

CWEs

CWE id Name
CWE-79 Improper Neutralization of Input During Web Page Generation ('Cross-site Scripting')

Credits

  • deprrous (reporter)

Affected packages (1)

Vulnerable version ranges and first patched releases as published by GitHub.

Ecosystem Package Vulnerable range First patched Vulnerable functions
npm @pdfme/schemas <= 5.5.8 5.5.9

References

cvelogic Threat Intelligence