moneyprinter-turbo is an automated, one-click AI short video production pipeline.
Given a topic, keyword, or user-supplied script, it orchestrates the entire production flow:
generating a multi-paragraph script, sourcing matching stock or AI video clips, synthesizing
natural voice-over narration, generating styled subtitles, mixing background music, and rendering
an exported MP4 in vertical (9:16), landscape (16:9), or square (1:1) format.
It provides four execution surfaces: an automated Agent runner (scripts/mpt_agent.py),
a Python CLI, a local Streamlit WebUI (port 8501), and a FastAPI backend (port 8080).
When to use this skill
- Generating finished short videos directly from a topic, keyword, or concept without existing footage.
- Automating faceless video creation for YouTube Shorts, Instagram Reels, or TikTok at scale.
- Generating voice-over narration using free, high-quality Edge TTS without an API key or account.
- Combining automated LLM scriptwriting with stock footage retrieval (Pexels, Pixabay) and BGM.
- Driving AI video generation models (MiniMax H3, Volcano Seedance, Wan, MuAPI) into coherent short videos.
- Running automated generation jobs from AI coding agents using the bundled
mpt_agent.pyhelper.
When NOT to use this skill
- Editing multi-take raw footage conversationally with coding agents → use
video-use. - Extracting and reframing clips from existing long-form video or podcasts → use
autoclip. - Code-first programmable video compositions with React and Remotion → use
video-production. - Affiliate marketing / e-commerce shopping shorts requiring strict platform compliance and disclosure → use
shopping-shorts. - Cinematic desktop product promo videos with 2.5D camera recipe cards → use
video-shotcraft.
Instructions
Step 1: Preflight and helper invocation
MoneyPrinterTurbo can be run directly via the bundled helper script scripts/mpt_agent.py.
Verify that Python 3.11+ and FFmpeg are available, then execute via uv:
uv run --no-project --python 3.11 python scripts/mpt_agent.py --subject "<video topic>"
The helper checks for an existing MoneyPrinterTurbo installation or installs it under ~/MoneyPrinterTurbo,
verifies dependencies, runs generation, and delivers the final MP4 path.
Step 2: Handle configuration and API keys
MoneyPrinterTurbo requires an LLM for scriptwriting and keyword extraction:
- Free/BYOK providers: OpenAI, Claude, Gemini, DeepSeek, Kimi/Moonshot, Qwen, Ollama.
- Voice-over (TTS): Defaults to free Edge TTS (no API key needed); Azure, ElevenLabs, Gemini also supported.
- Video materials: Pexels API key (free), Pixabay (free), local folder media, or AI video generation keys.
When running mpt_agent.py, if credentials are missing, it emits an input request (MPT_NEEDS_INPUT).
Provide keys via environment variables or configure config.toml:
export MPT_LLM_PROVIDER="openai"
export MPT_LLM_API_KEY="sk-..."
export MPT_PEXELS_API_KEY="..."
Step 3: Aspect ratio and language customization
Specify video orientation and language:
# Portrait 9:16 (Shorts/Reels/TikTok)
uv run --no-project --python 3.11 python scripts/mpt_agent.py \
--subject "History of Coffee" \
--aspect "9:16" \
--language "en"
# Landscape 16:9 (Standard YouTube)
uv run --no-project --python 3.11 python scripts/mpt_agent.py \
--subject "인공지능의 미래" \
--aspect "16:9" \
--language "ko"
Step 4: Inspect result
On successful completion, the runner outputs:
MPT_RESULT
VIDEO_FILE=/path/to/MoneyPrinterTurbo/storage/tasks/<task_id>/final-1.mp4
Verify the output MP4 exists and report the generated path and duration to the user.
Examples
Example 1: Create a 30-second science explainer short
User prompt:
Make an automated 9:16 short video about why the ocean is blue using stock footage and voiceover.
Workflow:
- Run the agent runner with subject "Why the ocean is blue":
uv run --no-project --python 3.11 python scripts/mpt_agent.py \ --subject "Why the ocean is blue" \ --aspect "9:16" - The pipeline writes the narration script, fetches Pexels ocean clips, generates Edge TTS audio, adds subtitles and ambient BGM.
- Deliver the rendered MP4 path to the user.
Best practices
- Leverage Edge TTS for zero-cost audio: Edge TTS offers high-quality natural voices in dozens of languages without subscription or API rate limits.
- Set specific topics: Narrow, concrete topics ("Why leaves change color in autumn") yield much more accurate stock clip matching than broad queries ("Nature").
- Review Seedance / AI video charges: When connecting paid AI video generation backends (Seedance, MiniMax H3), confirm billing limits with the user before triggering batch renders.
- Verify subtitle readability: Preview output videos to ensure text contrast and safe zones (avoiding bottom navigation UI on TikTok/Shorts) are respected.
References
- Supported Provider Matrix:
references/provider-matrix.md - Official repository: harry0703/MoneyPrinterTurbo
- Agent runner script:
scripts/mpt_agent.py