What problem does it solve? Cold emails often need a "not sure if this is you" referral angle, but naming the wrong colleague — or worse, naming the recipient back to themselves — destroys credibility. This Skill produces a safe, verified other_prospects variable (e.g. "Jeff Barg or Rob Cook") that slots into a referral CTA without ever naming the person being emailed. ## Core Features & Use Cases - Hard recipient exclusion: Enforces a never-name-the-recipient rule on every row using normalized name matching (first + last) plus LinkedIn URL matching, and refuses to run rows missing recipient_full_name. - Model judge with audit trail: A locked LLM prompt screens provider candidates, drops support staff and headline-contradicting titles (e.g. a "Co-founder" whose headline says wedding photographer), cleans names, and returns a dropped array explaining every rejection. - Still-there verification: A mandatory web check confirms each named person is still at the company before the variable ships, since provider job titles go stale. - Use Case: A sales team building a Clay table for outbound email adds this playbook so each contact row gets a phrase like "If this sits with Priya Ramanathan instead, happy to talk to them" — with the sentence disappearing entirely when the variable is empty. ## Quick Start Ask the AI to build the other_prospects variable for a prospect row by providing the domain, recipient full name, LinkedIn URL, and persona title list, following the exclusion and judging steps in this playbook.
