Write Outreach — Orchestrator

Communication — Horizontal. Ready-to-send outbound across email, LinkedIn, Twitter/X, platform proposals. Multi-agent strategy → draft → voice → critic → humanmaxxing pipeline.

Core Question: "If I removed the personalization, would this email still make sense? If yes, the personalization isn't working."

Why this skill exists, philosophy, methodology, principles, scope boundary, when NOT to use, what it pulls from elsewhere, history: references/playbook.md [PLAYBOOK].

Critical Gates — Read First

1. Mode + channel + target + proof are non-negotiable. Missing-Input Hard Blocks fire (BLOCK) when any of these are absent and not resolvable from artifacts. Signal can be missing with weak-signal flag; prior_touches required for touch 2+.
2. One ask per message, low-friction in touch 1. "Quick 30-minute call?" in touch 1 is too expensive for zero trust. Default to interest-question CTAs.
3. You > me ratio enforced structurally. First sentence (after salutation) does NOT start with "I" or "My." you/your-count must exceed I/we/our-count. Critic auto-fail when violated.
4. Humanmaxxing runs ONCE as terminal pass with protected_tokens. Running humanmaxxing twice strips specificity. Post-humanmaxxing Specificity regression check is automatic, not judgment — drops ≥2 or named entity/number absent → revert to critic-approved draft.
5. Never argue with a "no" in reply route. Breakup mode is default for firm not-interesteds. Critic auto-fails any reply that re-pitches after clear rejection, regardless of dim scores.

Quality Gate

Before delivering, the critic agent verifies (5 dimensions, 0-10 each):