Facebook Marketplace Safety Guard

Use this skill as a structured pre-send guard for every outbound Marketplace message and as a post-parse classifier for every inbound message captured by an inbox watcher or reply composer. The skill carries the pattern catalog for off-platform payment asks, deposit / holdback requests, premature identity exchange, shipping commitments outside the listing policy, and common scam tells.

It is detection-only. It never generates off-platform contact details, never proposes a payment rail outside the platform, and never suggests sharing identity information prematurely. It only flags those patterns in other text.

When to Use

Use this skill when:

• facebook-marketplace-message-sender is about to click Send on an outbound message — it must call this skill first.
• facebook-marketplace-buyer-inbox-watcher is classifying a newly-captured inbound message as part of its intent-parse step.
• facebook-marketplace-seller-reply-composer is preparing a reply and wants to verify the inbound the user is responding to wasn't a scam.
• The user pastes a seller / buyer message and asks "is this sketchy?"

Don't Use When