Skip to content

Security: trainingload-ai/trainingloadai

Security

SECURITY.md

Security Policy

Reporting a vulnerability

Do not report vulnerabilities, exploitation details, credentials, access tokens, database dumps, or athlete data in a public issue or discussion.

Use GitHub's private Report a vulnerability flow for this repository. If private vulnerability reporting is unavailable, contact the project maintainer through a private channel before sharing technical details. A useful report includes:

  • the affected release or commit;
  • the self-host mode and operating environment;
  • the security impact and required preconditions;
  • minimal reproduction steps or a proof of concept;
  • suggested remediation, when available.

Redact secrets and personal training data from screenshots, logs, diagnostics, and archives. Automated redaction reduces accidental exposure but does not make an artifact anonymous; inspect every artifact manually before sharing it.

Supported versions

Security fixes target supported signed releases. The current public repository is preparing its first signed self-host release, so source snapshots do not yet have a production support window. Once releases are published, supported versions and upgrade paths will be declared in release metadata and UPGRADE.md.

Self-host security boundary

The default self-host topology binds the Client to loopback and keeps Server and PostgreSQL on the private Compose network. It has one local Owner and no public signup. Exposing an instance to a LAN or the internet requires an operator-managed authenticated HTTPS reverse proxy, host hardening, backups, updates, and access controls.

Keep .env.selfhost, instance secrets, the PostgreSQL data directory, uploaded activity files, backups, diagnostics, and operation records private. Do not bypass signed release verification, migration guards, data-root checks, or lifecycle locks.

See SELF_HOSTING.md, SELF_HOST_OPERATIONS.md, and SELF_HOST_KNOWN_LIMITATIONS.md for the supported deployment boundary.

There aren't any published security advisories