Teen Safety Monitor

The monitoring bot is a server-side TypeScript enforcement service, not a decorative chat bot. It evaluates content before selected writes reach Discussions storage.

mt-4 text-5xl font-black tracking-[-.05em]

mt-5 text-sm leading-7 text-slate-500

Rules TEEN-001 through TEEN-030 describe the actual community contract and carry severity/default-action metadata.

Normalization, credential/phishing detection, privacy-risk signals, grooming/coercion patterns, malware/cyber-abuse intent, dangerous-content signals, link abuse and evasion patterns run locally.

Per-user event windows watch message velocity and repeated duplicate content instead of evaluating every post as an isolated string.

Optional classifier correlation

TEEN_SAFETY_CLASSIFIER_URL can add an independent classifier. Local enforcement still exists if that provider is unavailable.

Privacy-minimized auditing

Blocked/escalated decisions create private AETHER audit references containing a content hash, matched rule IDs, compact reasons, risk score and appeal state—not the full submitted message.

Discussion creation, discussion edits and replies are checked before URL-DB/AETHER persistence. /api/safety/moderate exposes the same engine to additional platform surfaces.

rounded-2xl border border-white/[.08] bg-white/[.025] p-5

mt-2 text-xs leading-6 text-slate-500

mt-7 inline-block rounded-xl bg-white px-4 py-3 text-xs font-black text-black