A complete go-to-market operating system for an early-stage B2B founder, in version control. Positioning docs, a lead pipeline with scoring scripts, cold-outreach sequences, sales playbooks, investor materials, content/PR strategy, and a set of AI-agent skills that automate the repetitive parts.
It was extracted from a real founder-led GTM motion (first design partner → first 10 customers) and generalized: every company-specific detail is a [PLACEHOLDER] you fill in, and every process doc is intact.
Built agent-first. The repo ships with CLAUDE.md (Claude Code), AGENTS.md (Codex and other agents), and vendored skills — so the fastest way to use it is to let your coding agent set it up for you and then run the motion with you.
1. Get the repo
git clone <your-fork-or-this-repo> my-gtm && cd my-gtm2. Open it with your agent — Claude Code, Codex, Cursor, or anything that reads repo instructions — and paste this:
Set up this GTM template for my company. Read TEMPLATE_GUIDE.md first, then
interview me one topic at a time (company + product, ICP segments, geography,
traction metrics, anchor customer, tool stack, sender domains). After each
answer, fill in the matching placeholders across the repo. When we're done,
grep for [A-Z_] bracket tokens and <!-- TODO --> comments and show me what's
still blank. Don't invent facts — ask.
Your agent walks you through a guided setup: it asks, you answer, it fills in the ~25 placeholder tokens everywhere they appear.
3. Sanity-check the strategy docs (the part only you can do):
positioning/ICP.md— does the segmentation match your market?positioning/messaging.md+pain-hypotheses.md— would your buyer nod along?qualification/scoring.md— do the signal weights make sense for your ICP?outreach/sequence-v1.md— would you reply to these emails?
Prefer manual setup? TEMPLATE_GUIDE.md has the full placeholder glossary for find-and-replace.
| Directory | What's in it |
|---|---|
positioning/ |
ICP segments, messaging, falsifiable pain hypotheses — read first |
pipeline/ |
Lead pool: raw → filtered → scored → enriched (CSVs + process) |
qualification/ |
0–100 ICP scoring rubric |
outreach/ |
Cold email sequence, warm-up checklist, draft templates |
playbooks/ |
Discovery, demo, design-partner agreement, objection handling |
investor/ |
Landscape, comps, fundraise plan |
investor-deck/ |
VC deck guide |
content/ |
Op-eds, press pitching, speaking targets, anchor report |
partnerships/ |
Distribution-partner playbooks + templates |
referrals/ |
Network graph — who introduced whom, who owes whom |
applications/ |
Accelerator / program application template |
trips/ |
Conference-trip planning template |
research/ |
Industry-standard / regulatory reference skeleton |
decision-log/ |
One file per meaningful decision — what changed and why |
metrics/ |
The weekly numbers that actually matter |
scripts/ |
Python (stdlib-only): filter → score → enrich → draft emails |
.agents/skills/ |
22 agent skills — auto-discovered, zero install (see below) |
- Position — ICP, messaging, pain hypotheses (
positioning/) - Build pipeline — export leads, filter and score them (
pipeline/build.md,scripts/) - Qualify — 0–100 rubric, work the top tier only (
qualification/scoring.md) - Warm up — sender domain + deliverability checklist (
outreach/warmup-checklist.md) - Reach out — 4-touch personalized sequence (
outreach/sequence-v1.md) - Discover — reply handling and discovery calls (
playbooks/discovery.md) - Convert — demo → design-partner agreement (
playbooks/) - Compound — referrals, content, PR (
referrals/,content/)
Measured weekly in metrics/weekly.md. Every strategy change gets a decision-log/ entry.
22 skills ship in .agents/skills/, the cross-agent Agent Skills standard location — Codex, Gemini CLI, Cursor, and other compliant agents discover them natively, and Claude Code picks them up via the symlinks in .claude/skills/. Open an agent session at the repo root and they're available:
| Group | What it does |
|---|---|
email-personalization |
The custom one. Lead row + buying signal → 90-word personalized first-touch email. Highest-leverage skill in the repo. |
gtm-* (10 skills) |
Pipeline mechanics — company/people/signal search, enrichment, contact filtering, outreach. Run /gtm-setup once to configure API keys. (from gtm-pipeline-skills) |
| Research & strategy (11 skills) | Lead intelligence, investor outreach + materials, market/deep research, article writing, brand voice, campaign planning, warm-intro graph ranking. Best with Exa/Firecrawl configured; works without. (from ECC) |
Windows: the
.claude/skills/entries are symlinks — enable symlinks (git config core.symlinks true+ Developer Mode) before cloning, or use the skills from.agents/skills/directly.
Each stage directory has a README with concrete options for running that part of the pipeline — from free/manual, to the agent skills in this repo, to paid tools — and what to deliberately skip. Start with pipeline/README.md and outreach/README.md. The whole motion runs on roughly $0–250/mo; this template optimizes for 30–50 sharply personalized touches a week, not 1,000 generic ones — the "What NOT to do" section in CLAUDE.md explains why.
- Decisions get logged. Changed the ICP, sequence, or scoring weights? Add a
decision-log/YYYY-MM-DD-<topic>.md. - Never commit contact data.
*_enriched.csv,*_with_emails.csv, and enrichment JSON are gitignored — keep it that way. - Send from a separate domain. Cold outreach goes from a dedicated subdomain, never your main domain.
- Docs over tools. Until ~10 customers, this repo plus a lightweight CRM is the whole stack.
MIT. The imported skill sets are also MIT — sources and full license texts in THIRD_PARTY_LICENSES.md. Use it, fork it, build your company on it.