Start

Install gtme and get to a receipt one of four ways, spending nothing until a page says so in plain words right before the command

What you'll learn

You have gtme installed, or are about to, and want a first result before reading anything else.

  • how to pick a starting page by what you have on hand
  • what each one needs, spends, and ends with
  • what a receipt is

Every page here ends in a receipt: a table of what a pipeline did, what it cost, and what it would have sent. Install gets the binary onto your machine. Then pick by what you have on hand.

Start is written for the person who runs campaigns, and every page here ends in a receipt that person can read. Two pages serve other readers and sit here so the on-ramp is in one place. Add a vendor is where the person adding a vendor begins, and For agents is where an agent begins. Someone building pipelines begins at Your CSV.

Start withNeedsSpendsEnds with
See it runnothing$0a receipt from recorded samples, then a second run that skips what the first one did
Your CSVa CSV and an Anthropic keycents, on the modelyour rows judged and written to a file
Your stackApollo, Harvest, Anthropic, and Instantly keys, plus a live Instantly campaignvendor credits and model tokens, gateda dry-run receipt, one armed run, then a re-run that sends nobody twice
Add a vendorjq and one request to a free public API$0a new adapter that verifies and simulates

These four are a menu, not a sequence. Each is one pipeline file you fetch, and every command is safe to re-run. Nothing sends and nothing spends until a page says so, in plain words, right before the command. We recommend See it run first, whatever you have on hand: it costs nothing and proves the install.

If an agent is going to do the typing, For agents has the two paste lines and the rules it follows. When you have a receipt and want to know what happened underneath, go to Concepts.