I take security and responsible handling of this project seriously. Please do not report security issues publicly via Issues or Pull Requests.

If you discover a vulnerability, security‑relevant behavior, or risk related to DNS / frAIme, please contact me directly:

[schltdns@gmail.com]

Provide as much detail as possible so I can reproduce and assess the issue.

1. I will acknowledge your report as quickly as possible.
2. I will investigate the issue and provide an assessment.
3. I will keep you informed about status and planned actions.
4. I will only publish security‑related fixes once the risk is resolved.

Thank you for helping keep this project secure.

• Input validation: all model outputs are checked against thresholds before logging.
• Auditability: hash-chaining (P8) ensures log integrity; manipulation is detectable.
• No persistent user data: the reference implementation stores no personal data.
• Multi-agent triangulation: reduces single-model manipulation risk.