Skip to content

Security: tbphp/gpt-load

SECURITY.md

Security Policy

Supported versions

Version line Status
2.0.x Pre-release security-supported candidate
1.4.x Maintained security support line

The repository's 2.0 candidate receives security fixes while its first public release is being prepared. This support status does not assert release readiness or public availability. No v2.0.0 tag, GitHub Release, public binary, or public container image is confirmed here; verify availability from actual public assets.

Reporting a vulnerability

Report suspected vulnerabilities through GitHub Private Vulnerability Reporting.

Please do not open a public issue for an undisclosed vulnerability. Use the private report so the maintainers can investigate and coordinate a fix before public disclosure.

Channel credential exposure boundary

Channel credentials and any derived secret fragments must not appear in data-plane response headers or bodies, request logs, or the request-log management API. Request-log attempts identify the selected credential only by its internal credential_id, together with the non-secret channel_id.

The authenticated Group credential list is the sole approved mask exception: it may return a safe identifier mask so an operator can identify a credential in the provider console. Plaintext reveal is a separate authenticated, explicitly invoked endpoint and must never be logged or returned by collection, health, usage, or request-log APIs.

Learn more about advisories related to tbphp/gpt-load in the GitHub Advisory Database