Quickstart

This guide takes you from a fresh install to your first extracted metadata in a few minutes. It uses the CSV extractor and the console sink, so you need no database or external service.

1. Install Meteor

brew install raystack/tap/meteor

See Installation for Docker, binary, and source options.

2. Create a sample data file

Make a small CSV file for Meteor to read:

echo "id,name,email" > users.csv

3. Create a recipe

A recipe is a YAML file that tells Meteor what to extract and where to send it. Generate one with the CLI:

meteor recipe init quickstart -n demo -e csv -s console > recipe.yaml

This creates a recipe named quickstart with the csv extractor and the console sink. The -n demo flag sets the scope, which becomes part of every URN Meteor generates.

Open recipe.yaml and set the path to your CSV file:

name: quickstart
version: v1beta1
source:
  name: csv
  scope: demo
  config:
    path: ./users.csv
sinks:
  - name: console

4. Validate the recipe

meteor lint recipe.yaml

Lint checks that the plugins exist and their configs are valid before you run anything.

5. Run it

meteor run recipe.yaml

The console sink prints each record as one JSON line, and Meteor ends with a run summary:

{"entity":{"urn":"urn:csv:demo:file:users.csv","type":"table","name":"users.csv","source":"csv","properties":{"columns":[{"name":"id"},{"name":"name"},{"name":"email"}]}}}

Each record contains an entity (the resource and its metadata) and, for sources that support it, edges (relationships like ownership and lineage).

Next steps

• Point Meteor at a real source. Pick one from the extractors list and swap the source block.
• Send metadata somewhere useful. Swap the console sink for Compass, Kafka, or another sink.
• Learn how recipes work in the recipe reference.
• Preview any recipe safely with meteor run recipe.yaml --dry-run --limit 10.