Skip to content

Security: freqkflag/neurolife

Security

SECURITY.md

Security Policy

Supported versions

Version Supported
0.1.x ✅ Active development

Reporting a vulnerability

Do not open public issues for security vulnerabilities.

Email or DM the maintainer with:

  1. Description of the issue and impact
  2. Steps to reproduce
  3. Affected components (API, web, mobile, sync)
  4. Suggested fix (if any)

We aim to acknowledge reports within 72 hours.

Security practices in this repo

  • Auth — JWT access + refresh tokens; bcrypt password hashing.
  • Encryption — Sensitive fields use @neurolife/encryption with ENCRYPTION_MASTER_KEY (rotate in production).
  • Documents — Stored in MinIO/S3; presigned URLs for upload/download.
  • AI privacy — PrivacyRouter in @neurolife/ai-core gates cloud providers; local/rule-based fallbacks when cloud keys are absent.
  • Audit — AuditService logs user actions to AuditLog (metadata stored as Prisma JSON).

Deployment checklist

  • Replace all change-me secrets in .env (JWT_*, ENCRYPTION_MASTER_KEY, MinIO credentials).
  • Restrict CORS_ORIGIN to your web origin.
  • Run Postgres and Redis on private networks; do not expose MinIO console publicly.
  • Enable TLS termination at your reverse proxy.
  • Review PrivacyMode per user before enabling CLOUD_ASSISTED AI.

Dependency updates

Run pnpm audit periodically. Security-related dependency bumps should be noted under ### Security in CHANGELOG.md.

There aren't any published security advisories