You built it. Nobody came.
Fix that.
A ten-agent growth team that installs into Claude Code. It finds the real constraint in your funnel, then builds the machine that gets your product found — keyword map to 50+ indexed pages — and cited, by ChatGPT, Perplexity, and Google's AI Overviews.
It diagnoses before it prescribes
Generic marketing agents start writing posts on day one. This one starts by finding what's actually broken — and refuses work that won't move the number.
34 visits a week, conversion fine → deploys the pSEO Architect. This is the kit's home turf.
Plenty of visits, 0.4% signup → SEO work refused. Routed to the CRO Reviewer instead.
Users leave in week two → no amount of pages fixes that. It tells you so.
The receipts
This pipeline ran on a real site before it was for sale. These are its numbers — not stock charts.
Brand and domain blurred — the site changed hands. The numbers didn't.
Meet the team
The star. Data source → templates → URLs → internal links → sitemap, implemented in your Next.js stack. With gates against thin content.
The new one. Scores any page on eight citability checks — quotable claims, original data, llms.txt, trust — and returns the fixes that get ChatGPT, Perplexity, and AI Overviews citing you.
The orchestrator. Diagnoses the funnel first, and refuses SEO work if your real problem is conversion.
Seed expansion → intent classification → clusters → a prioritized keyword map you can act on.
Long-form posts and comparison pages, sources cited. No filler intros — enforced.
Crawls your own site: meta, canonicals, sitemaps, redirect chains, broken links, structured data.
Builds and maintains the link graph. Varies anchors, catches orphan pages.
Message match, friction inventory, proof placement, pricing anchoring — on your landing and pricing pages.
Reads your GSC and analytics exports: striking-distance keywords, decay, refresh priorities.
X thread, LinkedIn, a Reddit version that respects each sub's etiquette, newsletter section.
Keyword map to indexed pages, in 4 commands
Seeds expanded, intent classified, clustered into page templates. Out: a CSV, prioritized.
The page system designed for your product: data source, templates, URL scheme, link graph.
Templates implemented, first batch generated. Thin pages rejected at the gate.
Every Monday: your analytics read, the story found, three actions. Forever.
Minimum unique data per page, thin-content detection. Feed it sparse data and it refuses to build — which is why Google keeps your pages. Fifty good pages beat five thousand junk ones.
Your buyers ask ChatGPT now
When an AI answers their question, it names two or three sources — and either you're one of them or your competitor is. This is the only Claude Code growth kit with a GEO team.
/growth:geo-audit grades any page against eight citability checks — quotable claims, original data, structure, freshness, trust — and tells you who gets cited today: you, a competitor, or a Reddit thread.
One idea per paragraph. Sentences an engine can quote verbatim. llms.txt, schema, and a stats page AI has to cite because the numbers exist nowhere else.
A scored E-E-A-T audit — the signals that make an answer engine comfortable naming you as the source. Citations follow trust.
Agencies now sell GEO as a monthly retainer. This is the kit version — you own it, it runs in your repo, and it re-checks live AI behavior instead of trusting last quarter's playbook.
What's in the box
agents — a lead that routes, specialists that ship. Each scoped to one job — including the only GEO agent in any Claude Code kit.
skills — the pSEO pipeline, GEO citability rulebook, scored E-E-A-T audit, keyword clustering, CRO checklist, GSC interpretation, Reddit & HN distribution rules, and more.
commands — /growth:diagnose through /growth:geo-audit.
offline scripts — cluster.py, link_audit.py. Local embeddings, no paid APIs required.
updates — SEO advice rots. The kit ships on a review cadence; updates are free.
Works with your stack
Is your site citable by AI? Check it — free
Paste a page URL and get its citability score — the same rubric the kit's GEO team uses to make pages that ChatGPT, Perplexity, and AI Overviews actually cite. The kit fixes the gaps; this shows you where they are.
Will this work for me?
$59 now. $69 after 10 more sales.
You built it. Nobody came. Fix that.
Stops your vibe-coded app from getting hacked. Defensive-only.
Every kit, one price. Save 20% over buying separately.
Save 20%Frequently Asked Questions
Do I need Claude Code? + −
Yes. The kit is a .claude/ folder — agents, skills, commands — that drops into your repo. If you built your product with Claude Code, you already have everything it needs.
Won't programmatic SEO get me penalized? + −
Thin pSEO will. That's why the build has quality gates: a minimum of unique data per page and thin-content detection. Feed it sparse data and it refuses to generate. Fifty good pages beat five thousand junk ones.
I'm not on Next.js. Is this useless to me? + −
The pSEO implementation patterns target Next.js (SSG/ISR). Everything else — keyword strategy, content, audits, distribution, analytics — is stack-agnostic. On Astro or Rails you'll adapt one skill, not twelve.
What's GEO, and do I actually need it? + −
Generative engine optimization — being the source that gets cited when ChatGPT, Perplexity, or Google's AI Overviews answer your buyer's question. A growing share of your ICP never sees a SERP anymore; the answer engine picks two or three sources and your competitor is one of them or you are. /growth:geo-audit scores any page on eight citability checks — quotable claims, original data, llms.txt, structure, trust — and returns the fixes in impact order. Agencies sell this as a monthly retainer; here it's a slash command.
Do I need paid API keys? + −
No. Every command works with zero paid APIs — that's the promise, and it's tested. If you already pay for DataForSEO or Serper, the kit detects your keys and silently upgrades from heuristic to data-backed mode. Optional power, never a requirement, never a nag.
How is this different from just asking Claude to "do SEO"? + −
Raw Claude gives you generic advice and enthusiastic filler posts. The kit encodes a method: constraint first, keyword map second, gated page systems third, weekly interpretation forever. It's the difference between a freelancer and a team with a process.
What if it doesn't work for me? + −
30 days, full refund, no questions. Email your receipt, get your money back.
Security Team
The team that stops your vibe-coded app from getting hacked. Defensive-only: it audits and hardens your own Next.js / Supabase / Stripe stack — before someone else tests it for you.