Let agents ship the pipeline

Agents write the pipeline, Deepline runs it. One CLI for waterfall enrichment across 105+ integrations, AI scoring, email validation, and sequencer push. Built for Claude Code, Codex, and Cursor.

Audience

Who this is for

Use Deepline to run GTM Plays from Claude Code, Cursor, or Codex. You do not need to move the work into a spreadsheet interface.

• You write scripts, not formulas. Your workflow belongs in a git repo, not a SaaS UI.
• You need agents to run enrichment autonomously, not click through a browser.
• You have provider API keys (Apollo, Crustdata, PDL) and want to use them directly.
• You need to inspect runtime limits and billing before you run a large input.

Workflows

5 workflows GTM engineers run with Deepline

Step 01: Waterfall lead enrichment

Start with 200 badge scans from a conference. Find a maintained waterfall Play and inspect its contract. Run the Play on a representative pilot.

deepline plays search "verified work email waterfall" --json
deepline plays describe prebuilt/name-and-domain-to-email-waterfall-batch --json
head -n 3 conference.csv > pilot.csv
deepline plays run prebuilt/name-and-domain-to-email-waterfall-batch --csv pilot.csv --watch

Step 02: Email verification and list hygiene

Validate each email before you send it to a sequencer. The validation contract must identify each supported email state for operator review.

deepline tools search "email validation" --json
# Claude Code authors email-validation.play.ts from the inspected tool contract.
deepline plays check email-validation.play.ts
deepline plays run --file email-validation.play.ts --csv contacts.csv --watch

Step 03: ICP scoring with AI

Enrich each lead with company data. Then use deeplineagent to score the lead against explicit ICP criteria. Use JSON mode to return fields that you can review and filter.

deepline plays search "company" --json
deepline tools search "company enrichment ICP scoring" --json
# Claude Code authors icp-scoring.play.ts from the inspected contracts.
deepline plays check icp-scoring.play.ts
head -n 3 leads.csv > pilot.csv
deepline plays run --file icp-scoring.play.ts --csv pilot.csv --watch

Step 04: Build a TAM list from scratch

Start with a list of target companies. Use one owned Play to find contacts, find and validate emails, and score the accepted records.

deepline plays search "company to contacts" --json
deepline plays describe prebuilt/company-to-contact --json
deepline plays describe prebuilt/name-and-domain-to-email-waterfall --json
deepline plays bootstrap company-people-email --from csv:target-companies.csv \
--people play:prebuilt/company-to-contact \
--email play:prebuilt/name-and-domain-to-email-waterfall \
--limit 500 --out tam-builder.play.ts
# Complete every generated TODO mapping, select the people output rows field,
# and remove the scaffold's guard throws before checking or running it.
deepline plays check tam-builder.play.ts
deepline plays run --file tam-builder.play.ts --limit 3 --watch

Step 05: Sequence automation

Add an idempotent sequencer step to the owned Play. Review a dry-run plan before the Play writes accepted contacts to the sequencer.

# Add a dry-run plan and idempotent sequencer stage to the owned Play.
deepline plays check tam-builder.play.ts
deepline plays run --file tam-builder.play.ts --limit 500 --watch

Step 06: Slack approval workflow

Deploy workflows that send you a Slack message with results, let you edit and approve, then continue to the next step, all in one flow.

Claude Code authors a webhook-bound Play with an explicit review boundary. Its definePlay options include an authenticated trigger:

{
  webhook: {
    auth: {
      type: 'standard-webhooks',
      headerFamily: 'standard',
      signingSecrets: ['LEAD_REVIEW_WEBHOOK_SECRET'],
      toleranceSeconds: 300,
    },
  },
}

The sender signs the exact body with Standard Webhooks v1 headers and reuses the same webhook-id when retrying the same event. Store the Play-scoped signing secret before Claude Code checks, publishes, and verifies the Play:

deepline secrets set LEAD_REVIEW_WEBHOOK_SECRET --scope play --play lead-review
deepline plays check lead-review.play.ts
deepline plays publish lead-review.play.ts
deepline plays describe lead-review --json
deepline plays versions --name lead-review --json

Agent Orchestration

Agents write the pipeline, Deepline runs it

The Deepline GTM skill teaches Claude Code how to inspect tools and Plays. Describe the required output and constraints. The agent can author an owned Play, run a pilot, and review the output.

> "I have a CSV of 200 leads from a conference. Enrich emails
>  with waterfall, validate them, score against our ICP, and
>  push the top 50 to Instantly."

# Claude Code reads the Deepline skill and runs:

deepline plays search "conference leads verified email ICP score sequencer" --json

# Claude Code authors conference-outbound.play.ts from the inspected contracts.

deepline plays check conference-outbound.play.ts
head -n 3 conference.csv > pilot.csv
deepline plays run --file conference-outbound.play.ts --csv pilot.csv --watch

# After review, the same Play runs the full input:

deepline plays run --file conference-outbound.play.ts --csv conference.csv --watch

Codex and Cursor can use the same CLI commands. Any agent with shell access can inspect and run Deepline Plays.

CLI vs Spreadsheet

Why the CLI beats the spreadsheet

Configurations

Common GTM engineer setups