Setup & Customization Checklist

1. Trigger setup

This skill runs from a LinkedIn Engagement trigger monitoring the tracked influencer's profile, with a broad topic filter (the skill's Step 1 is the precise gate). Reference trigger instructions:

When a tracked influencer publishes a LinkedIn post about [your company],
follow the <Influencer Post — Engager Capture> skill.

The skill will:
1. Verify the post is genuinely focused on [your company] (not just a
   mention in a list)
2. Filter out your team, company pages, and non-ICP engagers
3. Run full lead scoring on ICP-fit companies not already in a later
   pipeline stage
4. Post a digest to [signals channel]

• One trigger per tracked influencer (or one trigger monitoring a list, if your platform supports it).
• The trigger fires per-engager as engagement streams in — this is why the skill re-scrapes the full post (Step 2) and dedups against prior runs. Expect multiple runs per post.

2. External tooling

• Apify account with credits — two actors are load-bearing:
  • apimaestro/linkedin-post-comments-replies-engagements-scraper-no-cookies (full engager list; the trigger's list is always a subset)
  • harvestapi/linkedin-profile-scraper (batch person→company resolution; never parse companies out of headlines) Verify both actors still exist and their output fields match Steps 2–3; Apify actors change hands and schemas occasionally.
• LinkedIn tools connected for the Step 1b post-like (skip that step if unavailable).
• Business search + company enrichment tools connected (free search first, paid enrichment only after the eyeball ICP check — this is the cost-control pattern).

3. Placeholders

• {{COMPANY}}, {{INTERNAL_DOMAINS}}, {{TEAM_MEMBERS}} — the team list must be full names as they appear on LinkedIn, including teammates whose profiles don't list your company as employer.
• {{ICP_SEGMENTS}} — spelled out concretely (segment names + the obvious exclusions). This gate is where most engagers should die.
• {{SIGNALS_CHANNEL}}, {{LEAD_SCORING_SKILL}}, {{TIER_SCALE}}.

4. Policy decisions

• Detection + scoring only — this skill creates no outreach. Routing scored leads to outreach belongs downstream; keep it that way.
• Reactors vs commenters: the skill processes both, but comment scraping is the reliable path; reactor coverage depends on the trigger. Decide if reactor-only engagers are worth scoring for you.
• Digest granularity: per-lead top-level messages + one final digest is the tested format. High-volume influencer posts (100+ engagers) can flood a channel — consider a dedicated channel per campaign.

5. Behavioral invariants (do not remove)

• Step 1 precise focus check — a passing mention in a listicle must stop the run.
• Always scrape the post directly; never trust the trigger's engager subset.
• Never extract company from a LinkedIn headline.
• Dedup at three levels: against prior runs on the same post, across engagers at the same company, and against later-stage pipeline accounts.
• Never downgrade a funnel stage; active-deal accounts get a "warm signal" flag instead of re-scoring.
• Engagement alone caps at the second-lowest tier.