Deepline Quickstart

Quick Start

npm install -g deepline
deepline -h

Use the current Plays workflow for quick demos. Do not use the retired CSV enrichment command or its flags.

Single record

1. Ask which field the user wants to enrich and what identifiers they have.

2. Search for a maintained Play and inspect its input and output contract:

   deepline search "<input you have -> result you need>" --type prebuilts
   deepline describe prebuilt/<play-from-search>

3. Confirm the user wants to spend credits, then pilot one real record:

   deepline plays run prebuilt/<play-from-search> \
     --input '{"linkedin_url":"https://www.linkedin.com/in/example/"}'

4. Report the run result and the exact next step. Do not claim an output file exists unless one was written.

CSV

Inspect the file and choose a batch Play whose input schema matches its columns. Run a representative pilot before the full file:

deepline search "<input you have -> result you need>" --type prebuilts
deepline describe prebuilt/<batch-play-from-search>
deepline plays check prebuilt/<batch-play-from-search>
WORKDIR="deepline/data/quickstart-pilot" && mkdir -p "$WORKDIR"
head -n 4 leads.csv > "$WORKDIR/leads-pilot.csv"
deepline plays run prebuilt/<batch-play-from-search> --csv "$WORKDIR/leads-pilot.csv"
deepline runs export <run-id> --out "$WORKDIR/leads-pilot-enriched.csv"

Wait for a terminal run result before exporting. Review row alignment and results before asking whether to process the rest of the file. Never substitute a scalar Play for a batch Play unless its contract explicitly accepts rows.