Skip to content

Security: stenlysayd/neuro-engine

Security

SECURITY.md

Security Policy

Reporting Vulnerabilities

We take the security of Neuro Engine seriously, particularly because the system handles sensitive API credentials and cryptographic operations.

Do not file public GitHub issues for security vulnerabilities or key exposure concerns.

Instead, please report security issues privately via email to stenlysayd@gmail.com with:

  • A detailed summary of the vulnerability.
  • Steps to reproduce or a minimal proof-of-concept.
  • Potential impact on key leakage, unauthorized execution, or data exposure.

You will receive an acknowledgment within 48 hours, along with a remediation timeline.


Security Architecture & Best Practices

Neuro Engine is engineered with multiple defensive layers:

  1. At-Rest Encryption:

    • All third-party API credentials stored in SQLite are encrypted using AES-256-CBC with a unique Initialization Vector (IV) per key.
    • The master key is never persisted in the database; it must be supplied at runtime via the MASTER_KEY environment variable.
  2. Secret Masking & Zero-Leak APIs:

    • The internal API and telemetry events never return raw secrets to the frontend.
    • Keys are masked (sk-ant-***...3a9f) and identified via cryptographically salted hashes/fingerprints.
  3. Rate Limiting & Lease Cooldown:

    • Keys experiencing 429 (Too Many Requests) or 401 (Unauthorized) errors are automatically placed on temporary quarantine cooldown to prevent cascade account suspensions.
  4. Environment Isolation:

    • Database files (data/*.db) and audio cache files (data/audio-cache/) are strictly gitignored to prevent accidental commits of local runtime data.

There aren't any published security advisories