skill

Consensus Send Email Guard

Pre-send email governance with weighted persona votes, hard-block detection, and rewrite support.

Problem it solves

Outbound AI email can create legal, brand, and trust risk in a single send. This skill places a decision firewall before delivery.

Decision workflow

  1. load policy + persona set
  2. collect persona votes (YES | NO | REWRITE)
  3. aggregate by reputation
  4. enforce hard-block checks
  5. emit APPROVE | BLOCK | REWRITE
  6. persist decision + updated persona set

Why teams adopt it

  • blocks risky claims before external side effects
  • provides rewrite path for fixable drafts
  • logs every run for compliance and audit
  • updates evaluator influence over time

SKILL.md perspective

Stable JSON contracts with idempotent behavior and board-native artifacts.

AI-SELF-IMPROVEMENT perspective

Communication quality stops being subjective and becomes a measurable governance loop with replayable outcomes.