# sample-finance-bookcall

**Influencer ID**: <PENDING>
**Agent ID**: PENDING   <!-- this is a sanitized teaching sample, not deployed -->
**Use case**: appointment-setting (book-the-call)
**Vertical**: finance / trading education
**Platform**: instagram
**Template**: principles
**Flavor**: book-the-call
**Description**: Sanitized reference sample — a DM appointment-setter for a high-ticket day-trading mentorship; the DM qualifies (time, journey, goals, income, warmth) and books a call, the call sells.

## What this is

A SANITIZED, generic teaching example derived from a production finance / book-the-call agent.
It exists so a Claude operator can learn the SDK *shape*, the challenger (statement-before-question)
rapport, the anti-guru positioning, the time-commitment qualification gate, the long-cycle
re-engagement logic, and the "salaried team, not commission" transparency principle — without any
real persona, license number, income figure, handle, or link. **Not a live agent. Not deployed.**

All real/creator-specific/sensitive data has been stripped (see `changelog.md`). Persona is a generic
first name ("Mitch"), no surname, no regulator license number, no income claims. Credibility is carried
by technique only: "verified by an independent regulator (details disclosed on the call)" stands in for
any specific number.

## Folded / omitted from the richer source

The source carried extra operational files beyond the standard cortex-ops shape. They were mapped down:
- `qualification.json`, `routing.json` → folded into `knowledge_base.md` as prose (qualification
  dimensions, the income routing gate, post-audio skip logic, disqualifiers).
- `followups.json` → folded into `knowledge_base.md` as the long-cycle re-engagement section
  (the cadence is conceptual here; real workflows are configured separately via triggers/workflows).
- `contact-limits-spec.json` → the silent-stop conditions are summarized in `system.md` (NO_RESPONSE
  section) and `objections.md`; the operational custom-property/contact-limit wiring is omitted
  (it's external evaluator + platform config, not SDK prompt content).
- A keyword `audio_then_agent` opener was generalized into `keywords.json` + a note; the real audio
  resource id and transcript were removed.
