Clay Claude Code Skill Pack
Four Claude Code skill files for building production-grade Clay tables that generate pipeline, not experiments. Built for VP Sales, founders, and GTM engineers who use Clay to run outbound at scale and need their AI assistant to give precise, implementable output instead of generic advice.
These skills encode the decision logic, enrichment sequences, prompt architecture, and export structures that take a Clay table from idea to a running outbound campaign.
What's inside
| File | What it does |
|---|---|
.agents/skills/SKILL.md | Orchestrator and routing logic |
.agents/skills/clay-table-architect.md | Design a complete Clay table structure from an ICP and use case. |
.agents/skills/clay-enrichment-waterfall.md | Build or audit multi-provider enrichment waterfalls for email, phone, and firmographic data. |
.agents/skills/clay-ai-column-prompts.md | Write Clay AI column prompts that produce consistent, usable output at scale. |
.agents/skills/clay-outbound-export.md | Prepare Clay tables for export to Apollo, Smartlead, Instantly, and HubSpot. |
.agents/skills/clay-enrichment-debugger.md | Diagnose and resolve data match failures in Clay. |
.agents/skills/clay-ai-research-agent.md | Build deep custom research agents within Clay prompts. |
.agents/skills/clay-credit-optimization.md | Optimize enrichment waterfalls for maximum cost efficiency. |
Prerequisites
- Claude Code installed and running
- Active Clay account (clay.com)
- At least one enrichment provider connected in Clay (Apollo, Hunter, Clearbit, or equivalent)
- A sequencing tool account if using the export skill (Apollo, Smartlead, or Instantly)
Installation
Cursor, Windsurf, or Generic AI IDE
- Clone the repo:
git clone https://github.com/forma-norden/clay-claude-code-skill-pack - Copy the
.agents/skills/directory into your project's.agents/skills/folder.
Claude Code
- Clone the repo:
git clone https://github.com/forma-norden/clay-claude-code-skill-pack - Copy the
.agents/skills/directory into your project's.claude/skills/folder.
Usage
Load a skill by telling Claude Code to read it before describing your task:
Read .agents/skills/clay-table-architect.md
I need a prospecting table for B2B SaaS companies, 50-300 employees,
US and UK. Persona is VP Sales. Exporting to Apollo. Budget is $150/month
for 2000 rows.Claude will read the skill and produce a complete, production-ready table structure with column names, data types, enrichment waterfall, ICP filter logic, and cost estimate.
Running tests
See tests/README.md for full instructions. Short version:
- Load the skill
- Run the test prompt from the relevant test file
- Check the output against the validation checklist
Who this is for
GTM engineers, RevOps leads, VP Sales, and founders at B2B companies with 50 to 500 employees who are building or consolidating their outbound infrastructure and want to reduce tool sprawl through better-engineered GTM systems.
Ecosystem connections
This repository is designed to work as part of the Forma Norden open-source GTM ecosystem.
Works with
| Repo | Relationship | When to use together |
|---|---|---|
signal-based-list-building-workflow | Upstream | Signal-qualified lists feed into Clay tables for enrichment |
cold-email-copy-playbook | Downstream | Enriched data from Clay powers personalized email copy |
cold-email-deliverability-playbook | Downstream | Email verification in Clay connects to deliverability checks |
n8n-gtm-workflow-pack | Parallel | n8n orchestrates Clay enrichment via webhooks and HTTP requests |
buying-window-signal-workflow | Upstream | Signal scoring determines which leads enter Clay for enrichment |
linkedin-claude-code-workflow | Downstream | Enriched company data feeds LinkedIn ad targeting and outbound |
outbound-personalization-playbook | Downstream (planned) | Clay AI columns generate personalization data for outreach |
list-building-complete-playbook | Upstream (planned) | Complete list building workflows feed into Clay for enrichment |
Suggested skill chains
- Signal to enriched outbound list:
buying-window-signal-workflow(score) >clay-claude-code-skill-pack/clay-table-architect(design table) >clay-enrichment-waterfall(enrich) >clay-outbound-export(push to sequencer) - AI-powered personalization pipeline:
clay-ai-research-agent(gather data) >clay-ai-column-prompts(analyze and score) >cold-email-copy-playbook/copy-atl-executive-messaging(write personalized email) - Credit-optimized batch processing:
clay-credit-optimization(plan budget) >clay-table-architect(design with conditionals) >clay-enrichment-waterfall(run waterfall) >clay-enrichment-debugger(troubleshoot if needed)