AttackCompass

Blog

Plex2Shell: chaining four CVEs into a media-server compromise

Four CVEs in Plex Media Server chain from unauthenticated SSRF through internal pivot to arbitrary file write and code execution. All four were added to the CISA KEV catalog on 20 Oct 2026. Fixed in Plex >= 1.43.3.10861.

KEV · Web apps

Four vulnerabilities in Plex Media Server were added to the CISA Known Exploited Vulnerabilities catalog on 20 Oct 2026. The chain, informally called Plex2Shell, reaches unauthenticated remote code execution starting from SSRF on the media metadata fetch endpoint. All four are fixed in Plex Media Server >= 1.43.3.10861.

CVE chain summary

StepCVEClassDescription
1CVE-2026-96652SSRFPlex's media metadata fetch endpoint follows attacker-supplied URLs without sufficient host validation; reaches internal network services
2CVE-2026-96654URI manipulation / path traversalCrafted playlist URI causes path traversal in the media library scan path, exposing internal file listing endpoints
3CVE-2026-96655SSRF (internal pivot)Using the internal endpoint access gained from step 1, attacker reaches Plex's local management API on 127.0.0.1
4CVE-2026-96656Arbitrary file write to RCELocal management API accepts a file-write request (plug-in installation or script path); attacker writes a payload to a location that Plex executes on next media scan or plug-in load

Attack chain

Loading diagram…

Exploitation prerequisites and constraints

  • CVE-2026-96652 requires the Plex metadata fetch endpoint to be reachable. This endpoint is used for fetching media artwork and metadata from the internet; it is active on any Plex instance with network access enabled (default).
  • No authentication is required to trigger the initial SSRF. Plex uses session tokens for its web UI but the metadata fetch path is not gated the same way.
  • The SSRF reach depends on Plex's network position. A Plex instance with access to AWS EC2 instance metadata (169.254.169.254), GCP metadata, internal Kubernetes services, or other internal APIs is a higher-value pivot.
  • CVE-2026-96656 (file write to RCE) requires the attacker to know a path that Plex will execute. Plug-in directories and media scan scripts are the documented targets. The exact path varies by OS and Plex installation configuration.
  • The chain can be shortened if the attacker already has network access to the Plex management port (32400 or custom) — steps 1 and 3 may be bypassed in that scenario.

Patch and detection

Patch: Upgrade to Plex Media Server >= 1.43.3.10861. Downloads at plex.tv/media-server-downloads. Plex can auto-update; verify the running version from the Plex dashboard or Preferences.xml.

Detection artifacts to hunt (on a host or SIEM):

ArtifactLocationWhat it indicates
Unusual outbound HTTP from Plex processFirewall/proxy logsSSRF in progress — destination is internal IP space
Plex process spawning child shellEDR / process auditCVE-2026-96656 file-write RCE succeeded
New files in Plex plug-in directoryFile integrity / AVPayload staged for execution
Requests to 127.0.0.1:32400 from Plex processHost-based network monitorSSRF pivot to local management API
Unexpected media scan activity (off-schedule)Plex logs (~/.config/Plex/Logs/)Triggered to execute written payload

Operator notes

If Plex Media Server is in scope and you find an unpatched instance (version < 1.43.3.10861):

  1. Confirm network position: can Plex reach internal services? Cloud metadata endpoints are the first check.
  2. Test CVE-2026-96652 for SSRF before attempting the full chain — a successful metadata fetch to an attacker-controlled host (with response logging) is clean evidence for the report.
  3. Document the full chain as a compound finding with all four CVEs referenced. CVSS scoring for chains typically reflects the worst-case combined severity; report the four individually and the combined impact separately.
  4. Capture the Plex version string and configuration file as evidence before testing.

Sources

← All posts