GHSA-JXJR-3G7G-3944
Vulnerability from github – Published: 2026-09-08 21:03 – Updated: 2026-09-08 21:03Summary
An embedded line terminator bypasses the requireWellFormed serializer check for element and
attribute names. The check was added to fix GHSA-w2rr-34g9-rvrj and GHSA-4w3w-2rp5-g8jm; a name whose
first line is well-formed slips past it and is serialized verbatim, so the characters after the line
terminator break out of the start/end tag or attribute. Callers who enabled requireWellFormed
specifically to neutralize those name-injection issues remain exposed.
Details
xmldom builds every grammar production through a shared regexp builder that compiles with the m
flag. The anchored full-string matcher used for element and attribute names, QName_exact =
reg('^', QName, '$'), therefore inherits m. When it is applied as QName_exact.test(name) against
an already-assembled node name, the m flag makes $ match at an interior line terminator, so the
matcher accepts any value in which at least one line is a valid QName; the other lines are never
constrained. A payload whose first line is a valid QName, followed by a line terminator and breakout
markup, is what yields a working injection.
The serializer emits the accepted name verbatim into element start/end tags and attribute names, so
the bytes after the line terminator break out of the intended syntactic position. The check is
reached whenever a caller serializes, with requireWellFormed: true, a node whose name was set
through programmatic DOM construction (createElement, createElementNS, createAttribute,
createAttributeNS) with attacker-influenced input.
Root Cause
- A shared regexp builder compiles anchored productions with the
mflag. ^…$undermare line anchors, not string anchors.- A full-string validator built on such a production (
.test()) accepts any string with one conforming line, so a line terminator followed by breakout markup passes.
The triggering line terminators are the ECMAScript LineTerminator set: U+000A, U+000D, U+2028, U+2029.
Proof of Concept
const { DOMImplementation, XMLSerializer } = require('@xmldom/xmldom');
// Element name carrying an embedded line terminator + breakout markup:
const doc = new DOMImplementation().createDocument(null, 'root', null);
const el = doc.createElement('a\n><script>alert(1)</script');
doc.documentElement.appendChild(el);
// Caller opted into well-formed serialization, expecting invalid names to be rejected:
console.log(new XMLSerializer().serializeToString(doc, { requireWellFormed: true }));
// Observed on the affected version: NO throw; the output contains the injected `><script>…`
// breakout, because the name's first line ("a") satisfies the m-anchored QName check.
// Expected: InvalidStateError (the name is not a valid XML QName).
// Control — a single-line invalid name IS correctly rejected, proving the check is active and
// that only the line terminator defeats it:
const ctrl = new DOMImplementation().createDocument(null, 'root', null);
ctrl.documentElement.appendChild(ctrl.createElement('a b'));
new XMLSerializer().serializeToString(ctrl, { requireWellFormed: true });
// => throws InvalidStateError: The element name "a b" is not a valid XML QName
Impact
- Bypass of a previously shipped security mitigation. Applications that adopted
requireWellFormed: truespecifically to neutralize GHSA-w2rr-34g9-rvrj / GHSA-4w3w-2rp5-g8jm remain exposed to element/attribute name injection. - XML / markup structure injection, and, where the serialized output is placed into an HTML context, downstream XSS.
Fix Applied
The anchored XML Name/QName validators used by the requireWellFormed serializer no
longer treat interior line terminators as satisfying the anchors, so a name is validated against the
whole string. A name containing a line terminator is rejected with InvalidStateError, closing the
bypass for element and attribute names. The default serialization path is unchanged.
⚠ Opt-in required. Protection is not automatic. Existing serialization calls remain vulnerable unless
{ requireWellFormed: true }is explicitly passed. Applications that serialize untrusted DOM content should audit allserializeToString()call sites and add it.
Proof of Concept - fixed path
const { DOMImplementation, XMLSerializer } = require('@xmldom/xmldom');
const doc = new DOMImplementation().createDocument(null, 'root', null);
const el = doc.createElement('a\n><script>alert(1)</script');
doc.documentElement.appendChild(el);
// Default path (require-well-formed off) — unchanged, still emits the name verbatim,
// so the `><script>…` bytes break out of the start tag:
new XMLSerializer().serializeToString(doc);
// Opted-in path — now rejected:
new XMLSerializer().serializeToString(doc, { requireWellFormed: true });
// throws InvalidStateError: The element name "a\n><script>alert(1)</script" is not a valid XML QName
Why the default stays verbatim
The W3C DOM Parsing require-well-formed flag defaults to false, and browser XMLSerializer emits
names verbatim when it is unset. Throwing unconditionally would be an unjustified breaking change, so
the check stays gated on the caller opting in with { requireWellFormed: true }.
Residual limitation
The default serialization path (no requireWellFormed) still emits names verbatim by design (above).
Names introduced through createElement / setAttribute are never validated at creation — those APIs
store the name unchecked by design — so the opt-in serializer check remains the only guard on that
path.
{
"affected": [
{
"package": {
"ecosystem": "npm",
"name": "@xmldom/xmldom"
},
"ranges": [
{
"events": [
{
"introduced": "0.9.11"
},
{
"fixed": "0.9.12"
}
],
"type": "ECOSYSTEM"
}
],
"versions": [
"0.9.11"
]
}
],
"aliases": [
"CVE-2026-83617"
],
"database_specific": {
"cwe_ids": [
"CWE-625",
"CWE-91"
],
"github_reviewed": true,
"github_reviewed_at": "2026-09-08T21:03:17Z",
"nvd_published_at": "2026-09-01T15:17:40Z",
"severity": "HIGH"
},
"details": "## Summary\n\nAn embedded line terminator bypasses the `requireWellFormed` serializer check for element and\nattribute names. The check was added to fix GHSA-w2rr-34g9-rvrj and GHSA-4w3w-2rp5-g8jm; a name whose\nfirst line is well-formed slips past it and is serialized verbatim, so the characters after the line\nterminator break out of the start/end tag or attribute. Callers who enabled `requireWellFormed`\nspecifically to neutralize those name-injection issues remain exposed.\n\n## Details\n\nxmldom builds every grammar production through a shared regexp builder that compiles with the `m`\nflag. The anchored full-string matcher used for element and attribute names, `QName_exact =\nreg(\u0027^\u0027, QName, \u0027$\u0027)`, therefore inherits `m`. When it is applied as `QName_exact.test(name)` against\nan already-assembled node name, the `m` flag makes `$` match at an interior line terminator, so the\nmatcher accepts any value in which **at least one line** is a valid `QName`; the other lines are never\nconstrained. A payload whose first line is a valid `QName`, followed by a line terminator and breakout\nmarkup, is what yields a working injection.\n\nThe serializer emits the accepted name verbatim into element start/end tags and attribute names, so\nthe bytes after the line terminator break out of the intended syntactic position. The check is\nreached whenever a caller serializes, with `requireWellFormed: true`, a node whose name was set\nthrough programmatic DOM construction (`createElement`, `createElementNS`, `createAttribute`,\n`createAttributeNS`) with attacker-influenced input.\n\n### Root Cause\n\n1. A shared regexp builder compiles anchored productions with the `m` flag.\n2. `^\u2026$` under `m` are line anchors, not string anchors.\n3. A full-string validator built on such a production (`.test()`) accepts any string with one\n conforming line, so a line terminator followed by breakout markup passes.\n\nThe triggering line terminators are the ECMAScript `LineTerminator` set: U+000A, U+000D, U+2028, U+2029.\n\n## Proof of Concept\n\n```js\nconst { DOMImplementation, XMLSerializer } = require(\u0027@xmldom/xmldom\u0027);\n\n// Element name carrying an embedded line terminator + breakout markup:\nconst doc = new DOMImplementation().createDocument(null, \u0027root\u0027, null);\nconst el = doc.createElement(\u0027a\\n\u003e\u003cscript\u003ealert(1)\u003c/script\u0027);\ndoc.documentElement.appendChild(el);\n\n// Caller opted into well-formed serialization, expecting invalid names to be rejected:\nconsole.log(new XMLSerializer().serializeToString(doc, { requireWellFormed: true }));\n// Observed on the affected version: NO throw; the output contains the injected `\u003e\u003cscript\u003e\u2026`\n// breakout, because the name\u0027s first line (\"a\") satisfies the m-anchored QName check.\n// Expected: InvalidStateError (the name is not a valid XML QName).\n\n// Control \u2014 a single-line invalid name IS correctly rejected, proving the check is active and\n// that only the line terminator defeats it:\nconst ctrl = new DOMImplementation().createDocument(null, \u0027root\u0027, null);\nctrl.documentElement.appendChild(ctrl.createElement(\u0027a b\u0027));\nnew XMLSerializer().serializeToString(ctrl, { requireWellFormed: true });\n// =\u003e throws InvalidStateError: The element name \"a b\" is not a valid XML QName\n```\n\n## Impact\n\n- **Bypass of a previously shipped security mitigation.** Applications that adopted\n `requireWellFormed: true` specifically to neutralize GHSA-w2rr-34g9-rvrj / GHSA-4w3w-2rp5-g8jm\n remain exposed to element/attribute name injection.\n- **XML / markup structure injection**, and, where the serialized output is placed into an HTML\n context, downstream XSS.\n\n## Fix Applied\n\nThe anchored XML `Name`/`QName` validators used by the `requireWellFormed` serializer no\nlonger treat interior line terminators as satisfying the anchors, so a name is validated against the\nwhole string. A name containing a line terminator is rejected with `InvalidStateError`, closing the\nbypass for element and attribute names. The default serialization path is unchanged.\n\n\u003e **\u26a0 Opt-in required.** Protection is not automatic. Existing serialization calls remain\n\u003e vulnerable unless `{ requireWellFormed: true }` is explicitly passed. Applications that\n\u003e serialize untrusted DOM content should audit all `serializeToString()` call sites and add it.\n\n### Proof of Concept - fixed path\n\n```js\nconst { DOMImplementation, XMLSerializer } = require(\u0027@xmldom/xmldom\u0027);\nconst doc = new DOMImplementation().createDocument(null, \u0027root\u0027, null);\nconst el = doc.createElement(\u0027a\\n\u003e\u003cscript\u003ealert(1)\u003c/script\u0027);\ndoc.documentElement.appendChild(el);\n\n// Default path (require-well-formed off) \u2014 unchanged, still emits the name verbatim,\n// so the `\u003e\u003cscript\u003e\u2026` bytes break out of the start tag:\nnew XMLSerializer().serializeToString(doc);\n\n// Opted-in path \u2014 now rejected:\nnew XMLSerializer().serializeToString(doc, { requireWellFormed: true });\n// throws InvalidStateError: The element name \"a\\n\u003e\u003cscript\u003ealert(1)\u003c/script\" is not a valid XML QName\n```\n\n### Why the default stays verbatim\n\nThe W3C DOM Parsing require-well-formed flag defaults to false, and browser `XMLSerializer` emits\nnames verbatim when it is unset. Throwing unconditionally would be an unjustified breaking change, so\nthe check stays gated on the caller opting in with `{ requireWellFormed: true }`.\n\n### Residual limitation\n\nThe default serialization path (no `requireWellFormed`) still emits names verbatim by design (above).\nNames introduced through `createElement` / `setAttribute` are never validated at creation \u2014 those APIs\nstore the name unchecked by design \u2014 so the opt-in serializer check remains the only guard on that\npath.",
"id": "GHSA-jxjr-3g7g-3944",
"modified": "2026-09-08T21:03:17Z",
"published": "2026-09-08T21:03:17Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/xmldom/xmldom/security/advisories/GHSA-jxjr-3g7g-3944"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-83617"
},
{
"type": "WEB",
"url": "https://github.com/xmldom/xmldom/pull/1071"
},
{
"type": "WEB",
"url": "https://github.com/xmldom/xmldom/commit/7b2ec67e1750daadd0bb06c92e875e726544a362"
},
{
"type": "PACKAGE",
"url": "https://github.com/xmldom/xmldom"
},
{
"type": "WEB",
"url": "https://github.com/xmldom/xmldom/releases/tag/0.9.12"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "xmldom: requireWellFormed element/attribute name validation is bypassable via an embedded line terminator"
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.