CreatorForge AI

An AI content-creation platform for creators, businesses, marketers, and agencies. The long-term goal is to generate faceless videos, scripts, voiceovers, captions, thumbnails, blog posts, and social content.

This repository is Phase 1: the project foundation. Authentication, dashboard, project management, category selection, an AI script-generation placeholder, the full database structure, and payment architecture placeholders (Paddle + crypto). Full video generation is intentionally not built yet.

---

Tech stack

• Next.js 15 (App Router) + TypeScript
• Tailwind CSS (dark/light ready via CSS variables)
• Supabase — PostgreSQL + Auth (email/password) + Row Level Security
• OpenAI integration point (placeholder engine for now)
• Paddle + crypto payment structure (placeholders)

Quick start

# 1. Install dependencies
npm install

# 2. Configure environment
cp .env.example .env.local
#   …then fill in your Supabase values (see docs/ENVIRONMENT.md)

# 3. Set up the database
#   Open Supabase → SQL Editor → paste & run supabase/schema.sql
#   (see docs/DATABASE.md)

# 4. Run the dev server
npm run dev
# open http://localhost:3000

Without Supabase keys the marketing pages still render, but auth and the dashboard require a configured Supabase project. The script generator works without any AI key (it uses a built-in placeholder engine).

Project structure

creatorforge-ai/
├── src/
│   ├── app/
│   │   ├── page.tsx                 # Landing page
│   │   ├── pricing/                 # Pricing page
│   │   ├── (auth)/                  # login / signup (shared layout)
│   │   ├── dashboard/               # Authenticated app
│   │   │   ├── layout.tsx           # Sidebar + topbar shell
│   │   │   ├── page.tsx             # Dashboard home
│   │   │   ├── projects/            # list / new / [id] detail + server actions
│   │   │   ├── generate/            # Script generator
│   │   │   ├── voice/               # Voice Studio
│   │   │   ├── scenes/              # Scene Builder + timeline + subtitles
│   │   │   ├── thumbnails/          # Thumbnail generator
│   │   │   ├── assets/              # Asset library
│   │   │   ├── render/              # Render queue
│   │   │   ├── billing/             # Plans + payment placeholders
│   │   │   └── settings/            # Profile settings
│   │   └── api/
│   │       ├── generate-script/     # POST generate, PUT save
│   │       ├── voice/{preview,generate}/
│   │       ├── scenes/  images/  thumbnails/  subtitles/  assets/  render/
│   │       └── webhooks/            # paddle + crypto (placeholders)
│   ├── components/                  # UI + dashboard + auth components
│   ├── lib/
│   │   ├── supabase/                # browser + server + middleware clients
│   │   ├── ai/                      # generate.ts, providers.ts (Claude), prompts.ts
│   │   ├── media/                   # media engine: types, audio, scenes, subtitles
│   │   │   └── providers/           # voice / image / video registries (env-driven)
│   │   ├── payments/                # paddle.ts + crypto.ts (placeholders)
│   │   ├── credits.ts               # credit balance + atomic deduction
│   │   ├── constants.ts             # categories + plans + tones/lengths
│   │   └── types.ts
│   └── middleware.ts                # session refresh + /dashboard guard
├── supabase/schema.sql              # full DB schema + RLS + seed
├── docs/                            # DATABASE / ENVIRONMENT / PAYMENTS guides
└── .env.example

Documentation

• docs/ENVIRONMENT.md — environment variable guide
• docs/DATABASE.md — database setup instructions
• docs/PAYMENTS.md — payment architecture & where API keys go

Phase 2 (done) — AI generation & credits

• Real AI script generation via Anthropic Claude (src/lib/ai/): generate.ts orchestrates, providers.ts calls Claude (claude-opus-4-8 by default, override with AI_MODEL), prompts.ts holds per-category templates with tone + length controls. Falls back to a free offline placeholder when no key is set.
• Credit usage tracking (src/lib/credits.ts + deduct_credits() SQL function): real generations cost 1 credit, deducted atomically and logged to credit_usage. The generator pre-checks the balance and returns 402 with an upgrade prompt when the user is out of credits.
• Project save workflow: generate → save to a project (records model + token usage).

Phase 3 (done) — AI media generation engine

Modular, provider-based media pipeline (see docs/MEDIA.md). Ships with placeholder engines so it runs end-to-end without external keys:

• Voice Studio — voice/language/accent + speed/pitch, preview, generate, save, download.
• Scene Builder — auto-splits the script into editable scenes (narration, visual description, image/video prompts, camera direction, transition, duration), a reorderable timeline, per-scene image generation, and subtitle (SRT/VTT) generate/edit/download.
• Thumbnail Generator — 16:9 thumbnails with editable titles + styles.
• Asset Library — search/filter/download/delete across all media.
• Render Queue — placeholder render jobs with simulated progress, logs, and retry.
• New tables: assets, voiceovers, thumbnails, subtitles, render_jobs, scene_assets, media_library (+ extended scenes). Provider interfaces in src/lib/media.

Phase 4 — Track A (done): Storage & production foundation

Generated media now persists to Supabase Storage instead of data URIs / external links (see docs/MEDIA.md → Storage). src/lib/media/storage.ts uploads to a per-user folder in the public media bucket and stores the durable URL on assets; deleting an asset removes the underlying file. Run supabase/migrations/0004_phase4_storage.sql to create the bucket + RLS. Deployable to Vercel — see docs/DEPLOY.md.

What's next (Phase 4 — remaining)

1. Track B — real provider integrations (ElevenLabs voice, Stability/OpenAI images, Replicate video).
2. Track C — final video rendering (compose scenes + audio + burned-in captions) via a job worker.
3. Track D — Paddle + crypto checkout/webhooks; grant credits + monthly allowances on payment.
4. Team workspaces and API access.