openclaw-gtm-skills

An OpenClaw plugin + skill collection for full-stack B2B outbound GTM. Turns domains into researched, scored, segmented, and outreach-ready prospects. A Clay.com alternative you run yourself.

Inspired by and partially based on extruct-ai/gtm-skills. See ATTRIBUTION.md.

---

What's Inside

Standalone skills (can be used independently):

• context-building — build/update a GTM context file (ICP, voice, proof, campaigns, DNC list)
• hypothesis-building — generate testable pain hypotheses from ICP + win cases
• email-prompt-building — craft self-contained email prompt templates for campaigns
• email-generation — run prompts per-contact from CSV → generate personalized emails
• list-segmentation — segment + tier companies by hypothesis fit + data richness
• gtm-hypothesis-scorer — RICE scoring for all GTM hypotheses in your DB

Plugin pipeline (at plugins/beton-gtm/):

• gtm-intake → gtm-prefilter → gtm-research → gtm-sales-org → gtm-signals → gtm-segment → gtm-outreach → gtm-deck
• Full end-to-end: domain list → researched companies → fit scores → outreach emails → sales decks

---

Requirements

• OpenClaw installed and running
• Supabase project (for the pipeline database)
• Firecrawl — self-hosted or cloud
• Attio API key (optional — only for CRM sync in gtm-segment)
• seqd (optional — for outreach enrollment in gtm-outreach)

---

Standalone Skills Setup

Copy skills to your workspace:

cp -r skills/* ~/.openclaw/workspace/skills/

Each skill has a SKILL.md explaining when to use it and what it does. Example workflow:

1. context-building — capture product, ICP, voice, proof points
2. hypothesis-building — define pain hypotheses for a vertical
3. email-prompt-building — build a campaign-specific email prompt
4. email-generation — run the prompt against a CSV of contacts

See individual skill READMEs for details.

---

Plugin Pipeline Setup

1. Apply the database schema

supabase db push --workdir ./plugins/beton-gtm/supabase --password <your-db-password>

2. Configure the plugin

Edit plugins/beton-gtm/openclaw.plugin.json and fill in your values:

{
  "config": {
    "supabaseUrl": "https://yourproject.supabase.co",
    "supabaseKey": "your-service-role-key",
    "attioApiKey": "optional",
    "firecrawlUrl": "http://localhost:3002",
    "seqdApiUrl": "optional"
  }
}

3. Install in OpenClaw

Copy the plugin directory to your workspace:

cp -r plugins/beton-gtm ~/.openclaw/workspace/plugins/

4. Run the pipeline

Start with a CSV of domains:

cd plugins/beton-gtm/scripts
python3 run_intake.py --input domains.csv
python3 run_prefilter.py --limit 50000 --workers 20
python3 run_research_combined.py --limit 1000 --workers 5
python3 run_segment.py

---

Database Schema

Full Supabase schema in plugins/beton-gtm/supabase/migrations/. Tables:

• companies — core entity, tracks research_status through the pipeline
• company_classification — GTM motion, vertical, pricing model, B2B/SaaS flags
• company_firmographics — headcount, funding stage
• company_tech_stack — detected tools (CRM, analytics, etc.)
• company_sales_org — sales/revops headcount, hiring signals
• company_social — LinkedIn URL
• segments — your ICP segments with pain hypotheses + value props
• company_segments — company ↔ segment scoring (fit_score, fit_tier)
• value_hypotheses — segment-level pain hypotheses (generated by hypothesis-building)
• gtm_experiments — HADI cycle tracking for outreach experiments
• signals — buying signals per company
• contacts — people found for outreach
• outreach — drafted sequences, review status
• results — outcomes (replied, bounced, etc.)

---

Database Conventions

• Flat relational tables with foreign keys — no JSONB for structured data
• All schema changes via supabase migration new — never raw SQL
• Migration files are the source of truth

---

Recent Updates (last 2 weeks)

• Email writing rules locked in: no caps, capitalize given names, -> not →, "Sent from Superhuman" signature
• Subject line rule: name the problem/direction, NOT the company name
• New skills: gtm-hypothesis-scorer (RICE scoring), daily-closer (Sunsama EOD ritual)
• Hypothesis gen now filters to B2B SaaS only (b2b=true AND saas=true)
• Combined research + sales org into one script (run_research_combined.py)
• Value hypothesis generator added
• Vertical mismatch is now a hard disqualifier in segment scoring
• 30-industry segment taxonomy + updated VERTICAL_MAP
• Pricing teardown skill added with n8n hero images

---

Attribution

Based on extruct-ai/gtm-skills. See ATTRIBUTION.md.

---

License

GNU Affero General Public License v3.0

---

About

openclaw-gtm-skills is part of Beton, open-source revenue intelligence. See the dedicated page at getbeton.ai/oss-tools/openclaw-gtm-skills for an overview of the standalone skills and the end-to-end pipeline.

Other open-source tools we maintain:

• Inspector — the flagship signal-discovery agent
• DryFit — synthetic analytics datasets for agent benchmarking
• seqd — self-hosted email sequencer

Browse all open-source tools by Beton →