Setup & Customization Checklist

1. Senders and channels

• Fill {{SENDER_ROSTER}} — every person who can send, with which channels each has connected (email, LinkedIn). A sender without a healthy LinkedIn connection cannot carry LinkedIn steps; the skill stops rather than improvises.
• Assign {{AWARENESS_SENDER}} — a senior exec (founder/CxO) with an active LinkedIn presence. Their Stage 1 job is blank connection requests + one casual DM to decision makers: no pitch, no product mention. If no exec posts content regularly, skip Stage 1 rather than faking it.
• Assign {{HIGH_TIER_SENDER}} and {{LOW_TIER_SENDER}} per your {{TIER_MODEL}}. The point of the split is protecting the senior sender's reply rates and calendar — keep the boundary strict in both directions.
• Connect {{VOICE_SOURCE}} — per-sender voice guides and/or a library of past approved messages. If you have neither yet, run the first few sequences fully manually and save the approved versions; the skill drafts from approved copy, never from generic templates.

2. Placeholders

• {{CRM}}, {{SENDER_ROSTER}}, {{AWARENESS_SENDER}}, {{HIGH_TIER_SENDER}}, {{LOW_TIER_SENDER}}, {{TIER_MODEL}}, {{APPROVAL_OWNER}}, {{MULTITHREAD_THRESHOLD}}, {{LARGE_COMPANY_THRESHOLD}}, {{VOICE_SOURCE}}, {{RESOURCE_LIBRARY}}, {{AUTO_APPROVE_CARVEOUT}} — all filled.
• Map {{TIER_MODEL}} to your actual tiering scheme. The skill references tier semantics (high vs low), not names — any two-band split works.
• Build {{RESOURCE_LIBRARY}} as one page or doc: booking links, demo video, trust/security docs, DPA, one-pagers. It is the single source of truth — the skill never hardcodes a URL, so a moved link only needs fixing in one place.

3. Policy decisions

• Default {{AUTO_APPROVE_CARVEOUT}} to none. Everything queues for {{APPROVAL_OWNER}}. Only after weeks of clean approvals should you consider the narrow carveout in Step 4 (signed-up user, lifecycle trigger, ≤4 steps, not a senior exec at a large company) — and even then, all four conditions, no partial credit.
• The executive hard rule is not configurable downward. Sequences to C-level/VP/Head-of/Director contacts at companies above {{LARGE_COMPANY_THRESHOLD}} always require human approval, no matter what trigger fired. You may tighten the threshold, never loosen it.
• Set {{MULTITHREAD_THRESHOLD}} (default >200 employees). Below it, single-thread; above it, an exec multi-thread from {{AWARENESS_SENDER}} spawns automatically with every primary sequence — and always queues for approval.
• Decide whether low-tier accounts may auto-send at all. If your brand risk tolerance is low, queue everything — the skill works identically, just with more human clicks.
• Confirm the LinkedIn-active rule (10,000+ followers or visibly active → LinkedIn-only by default) matches your motion.

4. Behavioral invariants (do not remove)

• No hook, no send — generic "stay top of mind" requests get pushed back, not drafted.
• Never mention the signal directly — signals inform the angle, they are never quoted at the prospect.
• Scenario always wins over channel.
• Prior-relationship check before every pre-conversion play — replied outreach, past meetings, and closed-lost deals make it warm; unanswered sequences and anonymous visits do not.
• LinkedIn connection requests are always blank.
• Duplicate check before every new sequence — edit or archive, never stack.
• Written outreach rationale on every staged sequence — the approver should never have to reconstruct the "why."
• Plain-text URLs only in outreach copy — outreach channels don't render markdown.
• No attachments, ever — links from {{RESOURCE_LIBRARY}} instead.
• Approved messages get saved back to {{VOICE_SOURCE}}; unapproved drafts never do.