MoneyPrinter — 1-Click Short Video Automation Hub

From idea to published YouTube Short — fully automated, 100% local, near-zero cost.

MoneyPrinter is a local-first automation platform that turns a single idea into a finished vertical video and publishes it to YouTube, Twitter/X, TikTok, and Instagram — all from one dashboard, with one click.

---

Why MoneyPrinter?

Most "AI video" tools charge $30–$100/month per seat, lock you into their cloud, and still need you to babysit every step. MoneyPrinter is different:

• Run everything on your own machine. No cloud bills, no per-render fees.
• No paid API required. Use local Ollama models, KittenTTS, and local Whisper — or plug in any OpenAI-compatible endpoint (9Router, LiteLLM, LM Studio) if you want more power.
• One click = full pipeline. Research → Script → Images → Voice → Subtitles → Video → Upload. No tab-switching.
• Multi-platform publishing. Compose once, push to YouTube Shorts, Twitter/X, TikTok, and Instagram automatically.
• You stay in control. Manual review gate before publish. Edit any step, regenerate any stage, keep every session resumable.

---

The 1-Click Pipeline

Pick a topic. Press Generate Short. That's it.

Every stage writes to a session folder, so if anything fails you can resume or regenerate from the broken step — no lost work.

---

Feature Tour

YouTube Shorts Workspace

• Custom Subject + Audio Language selectors
• Auto Build button writes the full script
• Live Script Editor with manual tweaks
• Draft CC from text + Real CC from final audio
• Media Engine gallery for image review and manual swaps
• Generation Progress panel with per-stage regenerate

Research & Ideas Chat

A chat workspace that researches trends, brainstorms video ideas, and pushes the winning idea straight into a YouTube session. No more copying between tabs.

Twitter/X Manager

• Add multiple X accounts
• Compose + schedule posts from the dashboard
• Pulls recent post history per account

Affiliate CRM

• Amazon product scraping
• Auto-generated affiliate pitches
• One-click push to Twitter

Runtime + LLM Settings

• Pick your LLM backend (Ollama, 9Router, OpenAI-compatible, custom proxy)
• Pick image / TTS / STT / search models independently
• Enable fallback to local so the app never blocks on a missing key
• Tweak FFmpeg CRF, browser profile, FireFox automation paths

---

Cost Comparison

You bring the hardware (a Windows PC + Python 3.12 + Firefox). We bring the pipeline.

---

Quick Start (Windows)

Use this path when you only want to run the app.

# 1. Clone & enter
git clone https://github.com/mad-agentic/MoneyPrinter-short-video.git
cd MoneyPrinter-short-video

# 2. Copy config
copy config.example.json config.json

# 3. Install everything (Python venv + npm)
setup.bat

# 4. Launch the hub
start_hub.bat

Open http://localhost:5174 and you are live.

start_hub.bat checks backend port 15001 and frontend port 5174 before launch. If another process owns either port, stop that process or change ports before starting the hub.

Need the full local setup guide? Read docs/PROJECT_USAGE_VN.md.

Backend API: http://127.0.0.1:15001 · Frontend UI: http://localhost:5174

---

Requirements

Minimum: 0 paid API keys. The app is wired to fall back to fully local models at every stage.

---

Full Setup Checklist

1. Install prerequisites:
   • Python 3.12
   • Node.js 18+
   • FFmpeg
   • ImageMagick 7.x
   • Firefox, if you want browser-based YouTube/X publishing
2. Copy config.example.json to config.json.
3. In config.json, set at least:
   • imagemagick_path
   • LLM provider settings, for example Ollama or OpenAI-compatible/9Router
   • image, TTS, STT, and search models if you use provider routing
4. Run setup.bat once to create the Python virtual environment and install frontend packages.
5. Run start_hub.bat for normal use.
6. Open http://localhost:5174.
7. In Settings, verify provider, model, voice, ImageMagick, FFmpeg, and browser paths before generating or publishing.

Manual review is the default publish mode. Auto publish should be enabled only after account paths, metadata, and output video are verified.

Manual debug commands:

# Backend only
cd src
..\venv\Scripts\python.exe -m uvicorn api.main:app --port 15001 --reload

# Frontend only
cd frontend
npm run dev -- --host 127.0.0.1 --port 5174

Do not commit local runtime files: config.json, .env, .mp/, browser profiles, generated media, venv/, or frontend/node_modules/.

---

Project Structure

src/
  api/             FastAPI backend (port 15001)
    main.py        /system endpoints
    youtube.py     /youtube video pipeline
    twitter.py     /twitter posting
    affiliate.py   /affiliate products
    session_manager.py
  classes/         YouTube / Twitter / AFM / Tts / Outreach / PostBridge
  providers/       9Router, Ollama, OpenAI-compatible adapters
frontend/          React 19 + Vite (port 5174)
docs/              Screenshots, skills, this GitPage site
.mp/sessions/      Per-video session files (auto-resume state)

---

Multi-Platform Publishing

Compose once. Publish to:

• YouTube Shorts — full Selenium automation, scheduled, manual-review gate
• Twitter/X — auto-tweet from the same content
• TikTok + Instagram — available through PostBridge when post_bridge.enabled, API key, platforms, and account IDs are configured.
• Affiliate links — Amazon product pitches auto-pushed to X

---

Why "Local-First" Matters

• Privacy. Your scripts, account logins, and unreleased videos never leave your machine.
• Cost. No per-render fees, no seat licenses, no surprise overage bills.
• Speed. No upload-to-cloud, no waiting in a render queue.
• Resilience. No "service temporarily unavailable" mid-pipeline.
• Composability. Swap in any local model, any proxy, any voice — it's just config.

---

Screenshots Index

---

License

MIT — bring your own models, your own proxies, your own workflows. Use freely.