A public curriculum and starter kit for building modern full-stack applications with AI agents — host-agnostic, database-agnostic, and optimized for Cursor, Claude, Codex, ChatGPT, and Gemini.
This repository is an open, generic distillation of team practices for AI-assisted full-stack development. It is maintained by WDIMcKenzie for the coding community.
What is not included: private product code, proprietary security runbooks,
internal host or cloud naming, customer data, or any commercial product-specific
configuration. Use the template and playbooks with your own {{ORG}} placeholders.
You only need one primary coding tool in the repo — Cursor as
the Engineer. You focus on orchestration and architecture; the specialist
team in team/ maps to the tools below.
| SuperApp role | Tool | Use for |
|---|---|---|
| Orchestrator, Architect, Assistant, Document designer | Claude | Plans, blueprints, HANDOFF, specs, long docs |
| Engineer | Cursor | Implementation, tests, git |
| Assistant 2 | ChatGPT | Short coding checks, bugs, quick review |
| Researcher | Gemini / NotebookLM | Web search, research digests |
| Creative | Claude, ChatGPT, or Gemini | Graphics, UI, copy — any of the three |
| Quality | ChatGPT + Cursor | Pre-merge acceptance |
Stages: Intake (Claude) → Research (Gemini/NotebookLM) → Blueprint (Claude) → Implement (Cursor) → Quick check (ChatGPT) → Design (optional) → Quality → Ship.
Full setup steps, how to add each tool, and the daily loop:
docs/agents/multi-tool-orchestrator-setup.md.
Start Orchestrator day one: paste team/ADAPTER-PROMPT.md into Claude.
Coming soon: a We'll Do It track on building your own models (training, fine-tuning, evaluation) — follow welldoit.solutions for updates.
For production apps we typically deploy on Google Cloud (GCP), AWS, or
Oracle Cloud — each has a playbook under playbooks/hosting/.
Pair compute with the database matrix in
docs/hosting/database-host-pairing.md.
Everything above gets you a working starter and the workflow to build on it yourself. If you'd rather have it set up, hosted, or governed for you — hosting choices, deploy gates, ongoing support — We'll Do It Solutions offers free tech support and consulting, including help getting into our programs:
- Phone: 833-WDI-4YOU (833-934-4968)
- Email: specialist@welldoit.solutions
- Web: welldoit.solutions — book a session or explore programs
- User-branch governance —
user/<handle>/workspacedefault, review before staging, deploy-operator gate (handbook §06) - Multi-agent orchestration — Orchestrator + specialist team (
team/) - Dual-agent workflow — Architect (plans) + Engineer (implements)
- Starter monorepo — Next.js + API + swappable databases (
template/) - Hosting playbooks — AWS, GCP, Oracle, GoDaddy, Bluehost, Hostinger
- Database playbooks — Supabase, MongoDB, Cloud SQL, RDS, Oracle, local Postgres
- AI setup tiers — Ranked paths from one-session bootstrap to enterprise IAM
One command, after cloning:
pnpm run init # wizard: checks tools, then walks setup — project folder, DB, AIThe wizard runs the same tool checks as pnpm run doctor (a separate read-only
version exists if you just want a status report without the interactive setup).
No pnpm install is required to run the wizard itself — it's plain Node with
zero dependencies, so it works straight after git clone.
- Read docs/00-start-here.md
- Run
pnpm run init— createsprojects/<your-app>/with the standard layout - Open
projects/<your-app>/AI-START-HERE.mdand paste team/ADAPTER-PROMPT.md into your AI - From the project folder:
pnpm run db:validatethenpnpm dev
Offline-friendly default: the wizard's default database is local Postgres via Docker Compose — no external account or network call required beyond installing packages. Supabase/Mongo are opt-in for when you want hosted parity.
| Tool | Required | Purpose |
|---|---|---|
| Git | Yes | Version control |
| Node.js 20+ | Yes | Runtime (Volta recommended) |
| pnpm | Yes | Monorepo packages |
| Docker Desktop | For local DB | Postgres / Mongo containers |
GitHub CLI (gh) |
Optional | PRs and repos from terminal |
| Supabase CLI | If you pick Supabase | Local Supabase stack |
The wizard opens official download pages when something is missing — nothing is hard-coded to your machine path.
Tell your AI a trigger phrase; it runs the matching script from prompts/triggers/INDEX.md:
| Say | Does | Live deploy? |
|---|---|---|
| create new app | node scripts/ops/new-app.mjs |
No |
| bootstrap github | github-bootstrap.sh --confirm |
No |
| check the stack | stack-check.sh |
No |
| run the stack | stack-run.sh |
Local only |
| push to github | git-push.sh |
No |
| setup hosting | hosting-setup.sh --host gcp |
No |
| deploy staging | deploy-staging.sh --confirm |
Only with --confirm |
| deploy production | deploy-prod.sh --confirm |
Only with --confirm |
Full guide: docs/setup/ai-triggers.md
| Path | Purpose |
|---|---|
| docs/ | Handbook, curriculum, database contract, agent guides |
| team/ | Orchestrator + specialists, STATE/HANDOFF, SOPs |
| template/ | Production-style monorepo starter |
| samples/ | Small focused examples |
| playbooks/ | Host and database setup guides |
| scripts/ | Init wizard, doctor, leak scan |
| projects/ | Your generated apps (gitignored) |
| prompts/starter/ | Copy-paste AI starter prompts |
| prompts/triggers/ | Say a phrase → run a script (create app, GitHub, deploy gates) |
| scripts/ops/ | Ops scripts wired to triggers |
Run before every public push:
chmod +x scripts/leak-scan.sh
./scripts/leak-scan.shSee docs/security/publication-checklist.md.
MIT — see LICENSE.
- SuperApp Dev Stack — curriculum, template, playbooks, and ops scripts by the WDIMcKenzie / We'll Do It Solutions development team (MIT).
- Team coordination pattern — adapted from myPKA / ICOR (Paperless Movement / Tom Solid).
- Cursor (Anysphere) — recommended IDE and implementation agent; you orchestrate and architect, Cursor helps execute in the repo.
- Orchestrator stack — Anthropic Claude (Orchestrator/Architect), OpenAI ChatGPT
(Assistant 2 / quick checks), Google Gemini and NotebookLM (research), Cursor
(Engineer) — see
docs/agents/multi-tool-orchestrator-setup.md.
The starter monorepo and docs build on work from these communities (non-exhaustive):
| Layer | Projects |
|---|---|
| Web | Next.js, React |
| API | Hono, Node.js |
| Data | Prisma, Zod, MongoDB driver |
| Tooling | pnpm, Turborepo, TypeScript |
| Local DB (optional) | Docker, Supabase CLI, Postgres |
| Pinning (optional) | Volta |
Database and host playbooks also reference official guidance from Supabase, Neon, PlanetScale, MongoDB Atlas, Google Cloud, Amazon Web Services, and Oracle Cloud, plus shared hosting providers where WordPress split stacks apply.
- We'll Do It Solutions — Backend For Frontend Success.