Dopia

View saved version →

---
name: find-prospects
argument-hint: <your product or company URL>
description: Turn a product URL into an evidence-backed prospect list. Derives the qualification criteria and the channels from the product itself instead of applying a fixed template, harvests dated public signals, and gates every row on a quoted, linked, dated piece of evidence. Produces a list plus a corpus of the buyer's own words. Use when someone wants to find early customers, design partners, or a first outbound list for a product. Does not write or send outreach.
---

# Find prospects

Most prospecting tools sell rows from a database: name, title, company, email.
Those rows carry no date, no reason to talk, and no way to check whether they are
still true. This skill produces a different row: **a person, a dated public thing
they said or did, a link to it, and why that makes now the moment.** Every claim in
the output can be clicked back to its source.

🔴 **The one standard everything else serves: every row must be someone who could
plausibly pay for this product.** Not someone interesting, not someone who
described the problem well, not a company that merely fits the shape — someone
whose money could end up in the operator's account. That is the entire value
delivered, so it is the entire basis for judging the output.

Two things follow from it that are easy to skip. **Could pay means could pay at
the subject's own prices and on its own terms** — someone the subject's
published limits exclude, or whose whole need its free option already covers,
is not a buyer however well they describe the problem. And **where the site
shows that customers arrive through someone else**, the person who sends them is
how that money arrives, so they are worth finding by the same standard — in a
group of their own, never mixed in with the buyers.

Two consequences run through every phase below. **A competitor is worth less than
nothing** — their probability is zero and their presence makes the operator
distrust every other row, so no quality of evidence rescues one. And **how likely
is not yours to decide** — a plausible buyer at low probability costs the
operator one message, so rank by likelihood and let them draw their own line
rather than cutting on their behalf.

The part that makes the output non-generic is **not** the searching. It is
deriving *what to qualify on* from *what the product is worth*. Any skill that
hardcodes a check only works for products that happen to share that check. Derive
it; do not inherit it.

## What this needs

Capabilities, not specific tools. Use whatever the host provides.

| Capability | Used for | If missing |
|---|---|---|
| Fetch a public web page | product teardown, company checks | the skill cannot run |
| Web search | finding channels and posts | degrade to known registries only |
| A browser the host provides | any page plain fetch cannot read, and every lane on a platform that carries people's identities — those are browser-first, not browser-as-fallback | **say in Phase 3 which derived lanes this kills and what is left**, then run the short version knowingly |
| Write a local file | the report | print the report instead |
| An artifact tool *(optional)* | publishing the list as a shareable page | stop after the local file |
| A CRM tool *(optional)* | writing qualified rows as records | stop after the report |

## Boundaries

- **Prospecting only.** Do not draft, send, connect, comment, follow, or message.
  The output is a list and a corpus. Judge this skill on list quality, never on
  reply rate.
- **Public sources.** Marketplace reviews, support forums, trade communities,
  company websites — pages a person can open without an account. That is where
  the evidence this skill needs actually lives.
- **Never handle credentials, never defeat a login wall.** Do not ask for, accept,
  store or type a password, and do not work around anything a site put in front
  of its content to keep people out. If a page is gated, take anothe

Preview of the first 4,000 characters. Read full instructions →

Referenced files: 5 — file references are metadata, not package contents.