Do not include any of the following in issues, pull requests, trace samples, or support bundles:

• X-Plex-Token values or Plex client identifiers;
• MediaVault credentials, cloud-provider cookies, or API keys;
• complete signed CDN URLs or redirect query values;
• private hostnames, addresses, filesystem paths, or raw library activity logs.

Trace mode records paths and query parameter names but not query values, cookies, request bodies, response bodies, client identifiers, or remote addresses. Review traces before sharing because endpoint order can still reveal library usage patterns.

plex-probe reports exclude the Plex origin and token, but intentionally retain Part file paths and rating keys for stability analysis. Treat raw reports as private and publish only sanitized field shapes or comparisons.

Plex must authenticate requests arriving from the Gateway container network. Do not add the Gateway address or Docker subnet to Plex Allowed Networks. At startup, plex-gateway sends a request with a fresh invalid Plex Token and starts only when Plex rejects it with 401 or 403. This prevents transparent proxy requests from inheriting Plex's network-level anonymous access. Restart the Gateway after changing Plex network authentication settings so the boundary is checked again.

The configured MediaVault origin receives the complete active client header set during direct-link resolution, including Plex authentication headers and cookies. Configure only an operator-controlled MediaVault service, keep that control path on a protected network, and use HTTPS whenever it crosses a trust boundary.

The health and metrics endpoints do not contain request labels or credentials, but metrics expose aggregate activity counters. Restrict them at the reverse proxy when those counters should not be public.

Do not open a public issue containing exploit details or credentials. Email [email protected] with the subject plex-gateway security and provide the smallest sanitized reproduction possible.