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-2026-100850 Overview
CVE-2026-100850 is a Server-Side Request Forgery (SSRF) and local file read vulnerability in AzuraCast versions before 0.23.8. The flaw resides in the AutoDJ remote playlist fetch logic at backend/src/Radio/AutoDJ/QueueBuilder.php, specifically the getMediaFromRemoteUrl() function. A low-privileged user holding the station Media permission can supply a playlist with source=remote_url and remote_type=playlist pointing at file:// URIs or internal HTTP endpoints. The backend passes the attacker-controlled URL to file_get_contents() without scheme allowlisting or private-IP filtering, exposing host files and internal service responses through the queue API [CWE-918].
Critical Impact
Authenticated station operators can read arbitrary files readable by the web container, including /etc/passwd and the application .env file, and reach internal network endpoints such as cloud metadata services.
Affected Products
- AzuraCast versions prior to 0.23.8
- AzuraCast Docker image (ships with PHP allow_url_fopen enabled by default)
- Deployments granting the station Media permission to untrusted users
Discovery Timeline
- 2026-09-27 - CVE-2026-100850 published to NVD
- 2026-09-30 - Last updated in NVD database
Technical Details for CVE-2026-100850
Vulnerability Analysis
AzuraCast's AutoDJ subsystem builds the broadcast queue by fetching remote playlists configured by station operators. The getMediaFromRemoteUrl() function in QueueBuilder.php passes the user-supplied remote_url directly to PHP's file_get_contents(). PHP's allow_url_fopen directive is enabled by default in the shipped Docker image, so the function transparently handles http://, https://, and file:// schemes.
No scheme allowlist exists, and the code performs no filtering of loopback, link-local, private RFC1918, or cloud metadata (169.254.169.254) addresses. Lines returned from the fetched resource are parsed as M3U or PLS entries and stored in StationQueue.autodj_custom_uri. The stored values are then returned by GET /api/station/{station_id}/queue to any caller holding the Broadcasting permission, creating a non-blind read primitive.
Root Cause
The root cause is missing input validation on the URL scheme and target host combined with reliance on PHP's URL-aware file wrapper. The absence of a private-address policy allows the HTTP client to reach internal services, and the lack of a scheme allowlist permits direct local file reads via file://.
Attack Vector
An authenticated user with the station Media permission creates or updates a playlist, sets source to remote_url and remote_type to playlist, and populates remote_url with a target such as file:///etc/passwd, file:///var/azuracast/www/.env, or http://169.254.169.254/latest/meta-data/. When AutoDJ next builds the queue, the backend fetches the target and stores response lines in autodj_custom_uri. The attacker, or any user with the Broadcasting permission, retrieves the disclosed content from the queue API. Full technical details are available in the GitHub Security Advisory GHSA-rrjx-wrhf-8v47 and the VulnCheck SSRF Advisory.
Detection Methods for CVE-2026-100850
Indicators of Compromise
- Playlist records with source=remote_url and remote_url values containing file://, localhost, 127.0.0.1, RFC1918 ranges, or 169.254.169.254
- StationQueue.autodj_custom_uri entries containing fragments of /etc/passwd, environment variables, or internal service responses
- Unexpected outbound HTTP requests from the AzuraCast web container to loopback or metadata endpoints
- Access log entries showing GET /api/station/{id}/queue immediately following playlist creation or edit events
Detection Strategies
- Inspect the AzuraCast database for playlist rows where remote_url does not resolve to an external, publicly routable host
- Audit application logs for AutoDJ fetch operations that returned non-media content types or textual data resembling system files
- Monitor container egress for connections to 127.0.0.0/8, 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16, and 169.254.0.0/16
Monitoring Recommendations
- Alert on creation or modification of playlists with remote_type=playlist by any non-administrator account
- Capture and review responses from /api/station/{station_id}/queue for file content signatures
- Track file_get_contents() call patterns via PHP error logs or runtime application self-protection tooling
How to Mitigate CVE-2026-100850
Immediate Actions Required
- Revoke the station Media permission from any user that does not require playlist authoring
- Restrict network egress from the AzuraCast container to the public internet only, blocking loopback, RFC1918, and cloud metadata ranges
- Rotate secrets stored in the application .env file if exposure is suspected
- Review the queue API history for evidence of file disclosure
Patch Information
No patched version was available at the time of CVE publication. Track the GitHub Security Advisory GHSA-rrjx-wrhf-8v47 for release information and upgrade to version 0.23.8 or later when published.
Workarounds
- Disable PHP allow_url_fopen in the container's php.ini to block URL-aware file wrappers, accepting that this breaks legitimate remote playlist functionality
- Place AzuraCast behind an egress proxy that enforces a scheme and destination allowlist
- Deny file://, phar://, and other non-HTTP schemes at the application reverse proxy layer where feasible
- Limit the station Media and Broadcasting permissions to trusted operators only
# Disable URL-aware file wrappers in the AzuraCast PHP configuration
# Edit /etc/php/php.ini (path varies by image) and set:
allow_url_fopen = Off
allow_url_include = Off
# Restart the AzuraCast PHP-FPM service after applying changes
docker compose restart web
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

