Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-82387 Overview
CVE-2026-82387 is a stored cross-site scripting (XSS) vulnerability in Apache Roller 6.1.5. The flaw resides in the media upload feature, which trusts the client-supplied content type and serves the stored file back with that same type. An authenticated user with media-upload rights can upload a file containing active script content and have it executed in a victim's browser on Roller's origin. Media uploads are disabled by default, so only installations that explicitly enable them are affected. The shipped type restrictions do not block active content once uploads are enabled. The issue is tracked under [CWE-79] and carries a CVSS v3.1 score of 5.4.
Critical Impact
An authenticated low-privilege user can store script content that executes in another user's browser session on the Roller origin, enabling session theft, administrative action abuse, or redirection.
Affected Products
- Apache Roller 6.1.5 (installations with media uploads enabled)
- Earlier 6.x releases that include the same media-upload trust behavior
- Fixed in Apache Roller 6.1.6
Discovery Timeline
- 2026-09-28 - CVE-2026-82387 published to the National Vulnerability Database
- 2026-09-29 - Last updated in NVD database
Technical Details for CVE-2026-82387
Vulnerability Analysis
Apache Roller is a Java-based, open-source blog server. The media upload component accepts files from users holding upload rights and persists them to the Roller instance. When a client later requests a stored media item, Roller serves it using the content type recorded at upload time. Because that content type originates from the uploader and is not validated against the actual file bytes, an attacker can submit an HTML or SVG payload while declaring any MIME type. The server returns the stored payload with an executable type such as text/html, and the victim's browser renders the script within Roller's origin.
Execution in the origin context means the script can read authenticated cookies not protected by HttpOnly, invoke Roller administrative endpoints as the viewing user, or stage further client-side attacks. The vulnerability requires an authenticated account with media-upload rights, which limits the attacker pool but does not restrict impact once a privileged victim opens the file.
Root Cause
The root cause is improper neutralization of user-controlled input during web page generation [CWE-79], combined with blind trust in the upload-supplied Content-Type header. The shipped allow list of file types does not inspect file contents, so active markup slips past the restriction once uploads are enabled. See the fix in the Apache Roller pull request #174.
Attack Vector
An attacker with media-upload rights authenticates to Roller and uploads a crafted file containing JavaScript, declaring a permitted content type. The attacker then shares the resulting media URL with a victim, often an administrator or another authenticated user. When the victim opens the link, Roller returns the file with the attacker-controlled type and the browser executes the embedded script. User interaction is required, and the scope changes because script runs in Roller's trusted origin against the victim's session.
Detection Methods for CVE-2026-82387
Indicators of Compromise
- Media files served from Roller with content types inconsistent with their file extension or binary signature.
- Uploads submitted with Content-Type: text/html, image/svg+xml, or application/xhtml+xml to Roller media endpoints.
- Outbound requests from authenticated administrator sessions to attacker-controlled domains shortly after opening a media URL.
- Unexpected creation, modification, or publication of blog entries from legitimate administrator accounts.
Detection Strategies
- Inspect web server access logs for POST requests to Roller media upload paths followed by GET requests on the stored artifacts.
- Compare the declared MIME type stored in Roller's media metadata against the magic bytes of the file on disk.
- Alert on media responses from Roller that return text/html or SVG with embedded <script> elements.
Monitoring Recommendations
- Audit the list of users with media-upload rights and flag newly granted upload privileges.
- Forward Roller application and reverse-proxy logs to a central analytics platform for correlation of upload and retrieval events.
- Monitor for repeated uploads of small files with mismatched extensions and content types from the same account.
How to Mitigate CVE-2026-82387
Immediate Actions Required
- Upgrade Apache Roller to version 6.1.6 or later, which derives the stored content type from file bytes and serves non-image media as downloads.
- If upgrading immediately is not possible, disable the media upload feature until the patched release is deployed.
- Revoke media-upload rights from any account that does not strictly require them.
- Review existing media storage for files whose declared type does not match their contents and remove any suspicious artifacts.
Patch Information
The fix is included in Apache Roller 6.1.6. The patched release derives the stored content type from file content rather than the upload-supplied header and serves non-image media with a download disposition. Review the upstream change in the Apache Roller pull request #174 and the coordinated disclosure on the Apache mailing list thread and the OpenWall OSS Security update.
Workarounds
- Disable media uploads in Roller configuration until the 6.1.6 upgrade is applied.
- Place Roller behind a reverse proxy that rewrites response Content-Type for media paths to safe values such as application/octet-stream for non-image types.
- Add a strict Content-Security-Policy response header that disallows inline scripts on Roller's origin.
- Serve user-uploaded media from a separate, cookie-less domain to isolate any executed script from the Roller session context.
# Configuration example
# Example nginx snippet to force safe handling of Roller media responses
location /roller/media/ {
proxy_pass http://roller_backend;
proxy_hide_header Content-Type;
add_header Content-Type application/octet-stream always;
add_header Content-Disposition "attachment" always;
add_header X-Content-Type-Options "nosniff" always;
add_header Content-Security-Policy "default-src 'none'; sandbox;" always;
}
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

