Skip to content

README.md

Latest commit

 

History

History
147 lines (103 loc) · 5.14 KB

README.md

File metadata and controls

147 lines (103 loc) · 5.14 KB
Raw

GTM Pipeline Skills

Ten Claude Code skills for building efficient B2B lead generation pipelines — from company discovery to enriched contacts ready for outreach. Using these get you Clay quality data for 10-20% of the price.

Each skill is a standalone Markdown instruction file that Claude reads and executes as an agent. Use them individually or chain them into a full pipeline.

We are using ONLY the best and latest data providers and enrichment services and are not associated or affiliated with any of them.

Quick Start

git clone https://github.com/Keinsaas/gtm-pipeline-skills
cd gtm-pipeline-skills
./install.sh

Then open a project in Claude Code and run:

/gtm-pipeline:setup

The setup skill walks you through API keys, PhantomBuster agent IDs (auto-looked-up via MCP if available), and your first client ICP — in under 10 minutes.

Skills

Skill What It Does
/gtm-pipeline:setup Onboarding — install, configure keys, set up first client
/gtm-pipeline:pipeline Orchestrator — plan and run the full pipeline end-to-end
/gtm-pipeline:company-search Build a company list (Sales Navigator, Parallel FindAll, Firecrawl, web)
/gtm-pipeline:company-enrichment Enrich companies with headcount, revenue, growth metrics + ICP scoring (0–100)
/gtm-pipeline:signal-search Find buying intent signals (funding, hiring, leadership). Score each signal 1–100
/gtm-pipeline:people-search Find contacts by role at target companies, or prospect by persona
/gtm-pipeline:contact-filter Rank and filter contacts before enrichment — saves credits
/gtm-pipeline:people-enrichment Enrich contacts with verified work email and phone
/gtm-pipeline:outreach Run LinkedIn connection requests and messages via PhantomBuster
/gtm-pipeline:demo Generate a demo lead list (~10 contacts + messages) from an ICP description

Pipeline Architecture

Two workflows depending on what you start with:

Company-First (you have a company list or bounded market):

company-search → company-enrichment → ICP scoring → signal-search → people-search → contact-filter → people-enrichment

Signal-First (discovering companies via buying intent):

signal-search (discovery) → company-enrichment → ICP scoring → people-search → contact-filter → people-enrichment

See docs/ARCHITECTURE.md for the full flow diagram with CSV handoffs and cost checkpoints.

API Keys Required

Copy .env.example to your env file and fill in:

Key Provider Cost Model
PIPE0_API_KEY pipe0.com Credits per task
FULLENRICH_API_KEY fullenrich.com Per enrichment
SERPAPI_API_KEY serpapi.com Per search
PARALLEL_API_KEY parallel.ai Per task (processor-based)
FIRECRAWL_API_KEY firecrawl.dev Per crawl / per page
APIFY_API_KEY apify.com Per actor run
OPENROUTER_API_KEY openrouter.ai Per token
PHANTOMBUSTER_API_KEY phantombuster.com Plan-based
BETTERCONTACT_API_KEY bettercontact.rocks Per enrichment (optional)

You don't need all of them. See /gtm-pipeline:setup — it only asks for what your use case requires.

Project Structure

Each project (your company, a client, a campaign) gets its own working directory:

{slug}-gtm/
├── context/
│   ├── profile.md          # What they sell, value prop, tone
│   ├── icp.md              # Job tiers, industry tiers, location tiers, size filter
│   └── provider_performance.md
├── csv/
│   ├── input/
│   │   └── companies_raw.csv
│   ├── intermediate/
│   │   ├── companies_enriched.csv
│   │   ├── companies_scored.csv
│   │   ├── signals.csv
│   │   ├── contacts_found.csv
│   │   └── contacts_filtered.csv
│   └── output/
│       └── contacts_enriched.csv
└── run_log.md

The setup skill creates this structure for you. See examples/sample-client-gtm/ for a fully filled-out reference.

Personal Configuration

Your API keys and PhantomBuster agent IDs stay on your machine:

skills/_shared/local.md     ← gitignored, created from local.example.md by install.sh
~/.env.gtm                  ← default API key location (or set GTM_ENV_PATH)

What Not to Put in This Repo

  • API keys or session cookies (use ~/.env.gtm or your env manager)
  • Contact lists, lead data, or CRM exports
  • Client names, domains, or commercial terms
  • PhantomBuster agent IDs (go in _shared/local.md, which is gitignored)

For Production Integration

The skills handle data logic (enrichment, scoring, filtering). For automated production pipelines — webhook triggers, n8n orchestration, monitoring, scheduled runs — that's a separate integration layer. The skills are designed to compose with any orchestrator.

Contributing

See CONTRIBUTING.md. Fork the repo, make changes on a branch, submit a PR to main.

The stable branch is the maintainer's production copy — contributions go to main.

License

MIT