You spent 6 months building one SaaS that nobody bought.
You followed every checklist. You shipped. The market shrugged. Now what — six more months on another single bet?
Now shipping — used to build 60 products in 6 months
The boilerplate that builds 60 products,
not one.
FactoryStarter is the complete operating system for solo developers running an AI-leveraged product factory. Stop building one SaaS. Start running a portfolio.
30-day money-back guarantee · 12 months of updates included · Single-seat license
$ ./scripts/new-product.sh "ColdEmailTracker" webapp cold-email-tracker [new-product] plan: name: ColdEmailTracker type: webapp slug: cold-email-tracker source: ./boilerplates/webapp target: ../cold-email-tracker [new-product] copying boilerplate (186 files)… [new-product] initializing STATE.md, briefs/, .git… [new-product] done. $ cd ../cold-email-tracker && claude --dangerously-skip-permissions > Read STATE.md and resume from "Next". [spec-writer] writing brief for cold-email-tracker… [designer] tokens.json applied, components.md ready… [senior-dev] migrating supabase, wiring stripe, building landing… [tester] 17 vitest specs, 3 playwright specs — all green ✓ [devops] deploy-staging.yml triggered — vercel preview live ℹ Product 7/60 ready to launch.
The single-product trap.
Three patterns we've all lived. FactoryStarter exists because the third one broke me first.
Every new side project means rebuilding auth, payments, deploy from scratch.
Different repos, different stacks, the same tedious 40 hours of plumbing. By the time you reach the actual feature, the spark is gone.
Claude Code is powerful, but you waste hours per session re-orienting it.
No persistence, no agents, no playbooks. Each new product means re-explaining the stack, the conventions, the goals — over, and over, and over.
How FactoryStarter works
One repo. Sixty products.
FactoryStarter is the assembly line — your job is to pick which products run on it.
01
10 specialized AI agents, ready to ship products.
spec-writer, designer, senior-dev, tester, reviewer, devops, tech-writer, marketer, support-bot, analyst — each one a focused system prompt with clear handoff contracts. They run inside Claude Code so you stay in the terminal.
02
3 production-ready boilerplates: web, CLI, MCP.
Next.js 14 + Supabase + Stripe webapp scaffold; Node CLI scaffold; an MCP server scaffold. RLS on every table, signed webhooks, tests on the way in. No first-day plumbing.
03
Background agents that monitor your portfolio 24/7.
kill-monitor flags products below their kill rules. seo-checker watches Google Search Console. dependency-bot lifts security patches. analyst surfaces anomalies in PostHog. They run on cron, not on demand.
04
Playbooks for every recurring decision.
Ship-day, distribution, pricing-iteration, kill-or-scale, customer-support — eight battle-tested checklists. Every product follows the same path. Every path stays consistent across the portfolio.
Everything you get for $149.
One transparent list. If anything is missing or surprising, email me before you buy.
Repository
- ·186 files, 64 unit tests, 8 GitHub Actions workflows
- ·TypeScript strict, conventional commits enforced
- ·CI green on day one — lint + typecheck + unit + e2e + build
10 agent system prompts
- ·spec-writer, designer, senior-dev, tester, reviewer
- ·devops, tech-writer, marketer, support-bot, analyst
- ·Hand-off contracts written between every agent
3 boilerplates
- ·webapp: Next.js 14 + Supabase + Stripe, RLS-by-default
- ·cli-tool: Node + Commander, distributable via npm
- ·mcp-server: Model Context Protocol scaffold for Claude tools
5 background agents
- ·kill-monitor — kills products below their thresholds
- ·analyst — surfaces PostHog anomalies in Slack
- ·seo-checker, pr-reviewer, dependency-bot
8 playbooks · 15 prompts · 6 templates
- ·Ship-day, distribution, pricing-iteration, kill-or-scale
- ·Distribution / copywriting / SEO prompts
- ·PRODUCT_BRIEF, STATE.md, COMMIT_CONVENTIONS, AGENTS.md, …
Documentation
- ·AUTONOMY_FIRST — the cardinal rule for autonomous Claude Code
- ·PERSISTENCE — STATE.md conventions across sessions
- ·ONBOARDING, MULTI_AGENT, KILL_SCALE, BACKGROUND_AGENTS
Buyer access
- ·Read access to the private factory-template-0 GitHub repo
- ·Updates included for 12 months ($49/yr after, optional)
- ·Private buyer-only Discord with Marouane
Setup tooling
- ·setup-supabase, setup-stripe, setup-vercel, setup-cloudflare
- ·new-product.sh — bootstrap product N+1 in under 60 seconds
- ·tmux-setup, lint-all, backup-claude-sessions scripts
Honest comparison: FactoryStarter vs ShipFast vs DIY.
ShipFast helps you ship one SaaS. FactoryStarter helps you ship a portfolio. Different bet.
| FactoryStarter | ShipFast | DIY | |
|---|---|---|---|
| Focus | Portfolio of products | Single SaaS | Whatever you build |
| Price | $149 lifetime | $269 lifetime | Free + 200h of your time |
| AI agents (system prompts) | 10 specialized | None | You write them |
| Background monitoring | 5 cron agents | None | Build yourself |
| Boilerplates included | 3 (web · CLI · MCP) | 1 (web) | 0 |
| Playbooks | 8 written | README only | Tribal knowledge |
| Time to product N+1 | ~7-14 days | ~30 days | ~60-90 days |
| Updates included | 12 months | Lifetime updates | N/A |
| Refund window | 30 days, no questions | 7 days | N/A |
Comparison reflects ShipFast pricing and scope as of 2026-Q2; the FactoryStarter team has no affiliation with ShipFast.
Built in public. Used in public.
FactoryStarter is the same repo I use to ship every product on my 60-product challenge. You can read the BUILD_REPORT, watch the daily commits, and see what broke before you buy.
/ commits since v1.0
200+
All conventional, all reviewable. No `WIP`, no force-push to main.
/ products on the roadmap
60 in 6 months
Public roadmap. Public kill-or-scale calls. Real numbers.
/ tests passing
64 → 200+
Every product inherits the baseline; every product adds its own.
“Most boilerplates teach you to ship one product. FactoryStarter is the first one I've seen that teaches you to ship a portfolio. The agent architecture alone is worth the price.”
One plan. One price. Lifetime.
No tiers. No upsell. No paradox of choice — just access.
$149USD · one-time · lifetime
Buy once. Use it for every product you ship, forever.
- Read access to the factory-template-0 private GitHub repo
- 10 specialized AI agent system prompts
- 3 production-ready boilerplates (webapp · CLI · MCP)
- 5 background-monitoring agents (cron-driven)
- 8 playbooks, 15 prompts, 6 templates
- Complete docs (AUTONOMY_FIRST, PERSISTENCE, ONBOARDING)
- 12 months of updates included ($49/yr optional after)
- Private Discord with Marouane
- 30-day money-back guarantee, no questions
Stripe checkout, USD. Refunds within 30 days, no questions, single email.
Guarantee
Ship a product in 14 days, or get a full refund.
If FactoryStarter doesn't help you ship your first product within 14 days of starting — or if anything inside it isn't what you expected — email me athello@factorystarter.devwithin 30 days and I'll refund you in full. No forms, no “why are you leaving” questionnaire, no hidden re-engagement attempt.
The guarantee is the contract. You shouldn't need to argue for your money back.
FAQ
The questions Marouane gets in his X DMs every week.
Is the code open source?
No. FactoryStarter is proprietary. Buyers receive read access to the private factory-template-0 GitHub repo for the lifetime of their license.
Can I use it for client work?
Yes. There are no restrictions on what you build with FactoryStarter — internal tools, client sites, side projects, your own SaaS portfolio. Single seat, though: each developer needs their own license.
What if Anthropic changes Claude Code?
Updates are included for 12 months. When Claude Code ships a breaking change, the agents and playbooks get updated and pushed to the repo — you pull and you're current. After 12 months, the optional $49/year keeps the updates flowing.
Do I need to know AI / LLMs?
No deep knowledge needed. The agents are pre-configured system prompts; you run them via Claude Code's --dangerously-skip-permissions mode. If you can read a markdown file and run a CLI, you have what you need.
Can I share access with my team?
The default is a single seat. Teams should email hello@factorystarter.dev for a multi-seat license — typically 3-5 seats for the price of two.
What stack do I need to know?
TypeScript, Next.js, Supabase basics, Stripe basics. If you've shipped one indie SaaS before, you have everything required. The agents handle the boring parts.
How do refunds work?
Email hello@factorystarter.dev within 30 days of purchase. Full refund, no questions, no forms. Your GitHub access is revoked the same day.
Will I need other paid services?
Free tiers cover M1-M3 for most builders: Vercel hobby, Supabase free, Resend 3,000 emails/mo, PostHog 1M events/mo. You'll only pay when a product breaks through.
How long until I ship my first product?
7-14 days following the playbook. The first run takes longer because you're learning the agent flow; product 2 is usually ~5 days, and product N stabilizes at ~3-4 days.
Is this for me if I'm not a senior dev?
Best fit is intermediate-to-senior — you should be comfortable shipping a Next.js app with Supabase + Stripe to production. If that sounds heavy, ShipFast is probably a better starting point.
Stop building one SaaS. Start running a portfolio.
$149 lifetime · 12 months of updates · 30-day money-back guarantee.
Get instant GitHub access, today.
Built by Marouane Oulabass — full-stack engineer running a 60-product challenge in 6 months.