If you discover a security vulnerability in NEXO Brain, please report it responsibly:

1. Do NOT open a public issue
2. Email security@nexo-brain.com with:
   • Description of the vulnerability
   • Steps to reproduce
   • Potential impact
3. You will receive a response within 48 hours
4. We will work with you to understand and fix the issue before any public disclosure

NEXO Brain includes built-in security:

• 4-layer memory poisoning defense — validates all memory inputs
• Secret redaction — auto-detects and redacts API keys, tokens, passwords before storage
• Quarantine queue — new facts must earn trust before becoming knowledge
• Local-only processing — all vectors computed on CPU, no cloud dependencies
• Filesystem-protected SQLite — credentials are stored in local SQLite; protect the database with filesystem permissions (chmod 600). For production secrets, use environment variables or a dedicated secret manager

The following are in scope:

• Memory injection/poisoning attacks
• Secret leakage through memory retrieval
• Authentication bypass in MCP tool calls
• Data exfiltration through plugin system

The following are out of scope:

• Vulnerabilities in upstream dependencies (report to them directly)
• Social engineering attacks
• Denial of service through excessive memory writes (rate limiting is the user's responsibility)