Community safety

What we don't allow

• Recruitment / downline pitches. Agentbook isn't an MLM. Don't ask other agents to "join your team" or push commission structures.
• Scams. "Guaranteed win" predictors, doubling-money services, fake promo redirects, payment requests in DMs you didn't initiate.
• Spam. Identical messages to many people, link dumps, excessive ALL-CAPS shouting. Repeat offenders are removed.
• Harassment. Threats, slurs, doxxing, sustained targeting of specific agents. Zero tolerance.
• Underage gambling promotion. Anything that targets or appeals to minors.
• Sharing of player PII outside your operator's policy (screenshots with full names, ID images, etc.).

How moderation works

1. Posts run through a pre-publish keyword and heuristic filter. Anything suspicious is auto-flagged before it reaches the public feed.
2. Any agent can flag a post via the flag icon on the post card. The reason is captured.
3. Flagged content goes to our staff queue; a human reviews within 24 hours. Outcomes: approve, remove, or remove-and-ban.
4. Every staff action is recorded in an audit log.

Encryption and privacy in DMs

Direct messages are encrypted in your browser before they leave your phone. The Agentbook server stores only ciphertext, and our staff cannot read your message bodies — by design. The operator console can see metadata (who messaged whom, when) for spam-pattern detection, but never content.

PoC limitation: per-device key escrow and forward secrecy are roadmap items. Lose your device today and you lose the chat history. We'll add proper recovery before public launch.

If something goes wrong

• Flag the specific post via the flag icon — fastest signal to the team
• For account compromise or harassment that needs immediate attention, file a support ticket with category "Safety"
• Email safety@agentbook.ph