Outbound Send Gate

0•

Enforces human-in-the-loop safeguards and strict validation gates whenever AI agents draft, stage, or interact with outbound communications.

Outbound Send Gate provides an incident-tested safeguard framework that strictly divides AI agent actions from operator authority: agents stage outbound communications, while human operators review and send them. Designed for web automation, email workflows, and form submissions, this skill prevents accidental message dispatch through seven rigorous process gates—including sent-count sentinels, DOM-level identity verification, clipboard integrity checks, and dynamic coordinate handling. It is essential for any automated clinical, business, or operational workflow where unvetted outbound communications or identity mismatches carry critical real-world risk.

Key Features

01Strict separation of agent draft/staging roles from human send authority
02DOM-level validation of 'From' identity and saved draft content prior to handoff
03Dynamic UI interaction rules preventing misclicks on send, minimize, or close controls
04Pre- and post-session sent-count sentinels across multiple send-as identities
05Automated detection of unsupported first-person claims requiring human verification
060 GitHub stars

Use Cases

01Filling out external web portals, job applications, or contact forms for manual final submission
02Staging batch outreach or email drafts in Gmail and webmail clients without accidental sends
03Managing multi-identity communication workflows to guarantee correct sender addresses