A multi-tenant Django + Next.js boilerplate. Auth, billing, roles, Stripe, i18n — already wired. Buy once, ship every SaaS idea you have for the rest of your career.
$ git clone git@github.com:you/launchkit.git my-saas
Cloning into 'my-saas'...
$ cd my-saas && cp .env.example .env
$ docker compose up -d && python manage.py seed_demo
[+] Running 4/4 — postgres, redis, api, worker
[seed] 1 demo tenant, 3 plans, 12 members, 5 tickets
$ open http://localhost:3000
→ Dashboard live. Stripe wired. Time to ship features.
▍Production-grade stack — already wired
app.launchasaas.dev/dashboard Auth · Sign in Settings · Members Settings · Subscription Three weeks of plumbing before a single feature line lands. Skip it.
Auth · JWT refresh · Password reset · Email verify · 2FA · Google OAuth
Tenants · Roles · Member invites · Permission system with wildcards
Stripe · Plans · Addons · Webhooks · Idempotency · Past-due grace
i18n · Audit logs · Rate limits · Error pages · Tests · CI · …
Not a starter template. Not a tutorial with TODOs. Full backend + frontend, audited end-to-end, every unsexy part already done — and tested.
Every request gets a TenantMiddleware-resolved tenant. ViewSets auto-scope queries. New models inherit TenantOwnedModel. You literally have to opt out of tenancy, not opt in.
# apps/your_feature/views.pyclass InvoiceViewSet( TenantViewSetMixin, # auto-scopes by request.tenant TenantPermissionMixin, # checks app.resource.action SubscriptionLimitMixin, # enforces plan quotas viewsets.ModelViewSet,): serializer_class = InvoiceSerializer queryset = Invoice.objects.all() # That's it. 5 lines. Multi-tenant + RBAC + plan-gated.Idempotent webhook handlers backed by a StripeEvent table. Credit purchases dedup on payment_intent_id. CreditBalance updates run inside select_for_update() — race conditions caught and tested.
# apps/subscriptions/tasks.py@shared_taskdef handle_credit_purchase_completed(event_id, session): with transaction.atomic(): balance = CreditBalance.objects.select_for_update() .get_or_create(tenant=tenant)[0] # Re-check inside the lock — Celery may retry. if already_processed(payment_intent_id): return # idempotent. nothing to do. balance.deposit(amount, metadata={...})JWT with rotation and blacklist. 2FA via email codes or TOTP. Google OAuth. Per-user lockouts after N failures. Hardened across multiple security + quality audits — every finding pinned by a regression test.
# apps/users/tests/test_security_regressions.pydef test_email_field_is_read_only_on_profile_patch(api): # Closes account-takeover via PATCH /auth/user/ response = api.patch("/auth/user/", {"email": "x@y.z"}) assert response.status_code == 200 assert User.objects.get().email != "x@y.z" # 16 of these. Every audit finding has its own test. CLAUDE.md maps every
load-bearing pattern. Five skill recipes cover the domains agents
(and humans) most often break. Works with any LLM-backed assistant.
/launchkit-permissions How to add a new permission, register it in the catalog, and run update_default_roles.
/launchkit-stripe-webhook Idempotency rules, dedup on payment_intent_id, transaction.atomic + select_for_update patterns.
/launchkit-add-app The full template for adding a new Django app with tenant scoping, ViewSet mixins and i18n.
/launchkit-i18n Backend gettext + frontend next-intl, with the auditor scripts that catch missing keys.
/launchkit-add-feature Frontend recipe: Server Component page → client component → API client → query keys → routes. # LaunchKit API — request pipeline
Every request flows through:
1. JWTAuthenticationMiddleware → request.user
2. TenantMiddleware → request.tenant
3. SubscriptionMiddleware → 403 if expired
4. TenantPermissionMixin → app.resource.action
5. ModuleAccessMixin → entitlements
6. SubscriptionLimitMixin → quota check
## Things to verify before editing
— New URL prefix that bypasses tenant?
Update both TENANT_MIDDLEWARE_EXCLUDED_PATHS
and SUBSCRIPTION_EXEMPT_PATHS.
— New Celery webhook handler?
Must be idempotent. Mark StripeEvent.processed=True.
Must catch + retry. Capture to Sentry.
…CLAUDE.md ready | Option A LaunchKit | Option B Other boilerplate | |
|---|---|---|
| Time to first dashboard | 5 min | 2–3 days |
| Multi-tenant by default | Yes | Maybe |
| Stripe webhook idempotency | Tested | Hopeful |
| Permission system with wildcards | Yes | Sometimes |
| Automated tests | 400 backend + 29 frontend | Demo only |
| i18n (EN + ES, both stacks) | 100% covered | Frontend only |
| AI-agent docs (CLAUDE.md + skills) | Yes | No |
| License | One-time, commercial ∞ projects on every tier | Subscription |
Three weeks of plumbing at a junior contractor's rate is roughly $4,800. Three weeks of your time is whatever opportunity cost you'd rather not put a number on.
No subscriptions. No per-seat creep. Pay once, own the source, ship as many SaaS projects as you can dream up.
one-time · Unlimited projects · Use commercially
Secure checkout · Repo access in 5 min
one-time · Unlimited projects · Use commercially
Secure checkout · Repo access in 5 min
One payment. Lifetime license. Build the SaaS your customers see — not the one you'll never finish.
Buy LaunchKit — from $249Secure checkout · Repo access in your inbox · Read the FAQ before buying