MoneyPrinterTurbo
Turns a topic into a finished vertical short video on its own host, writing the script, sourcing stock footage, narrating it, and burning in subtitles, then returns the render as a pollable task.
What it provides
Connect an HTTPS inference API. Model weights stay on CPU/GPU infrastructure and are never installed inside the 4 GB bot.
Connection requirements
Run `uv run python main.py` outside the bot, which serves the API on port 8080, and publish it behind an authenticated HTTPS reverse proxy — upstream ships its API auth disabled, so without that proxy anyone who finds the URL can spend your render time and provider credits. Keep Bearer selected and paste the proxy token below. Give that host several spare CPU cores: rendering is real video transcoding, and a single core takes about half an hour for a one-minute 1080x1920 clip, so keep it off the machine running your bot. Generation is asynchronous — `post /api/v1/videos` returns a task_id straight away, then poll `get /api/v1/tasks/TASK_ID` until progress reaches 100. The finished MP4 stays on that host: this connector prints responses rather than saving them and caps a printed response at 5 MB, so a rendered video is far too large to pull back through the bot. Have the agent hand over the server’s own download URL instead. `video_subject` is the only required field in the request, but the host needs its config.toml filled in first: the LLM key, the text-to-speech key, and a stock-footage key, because `video_source` defaults to Pexels and material sourcing fails partway through a task when `pexels_api_keys` is empty. Point it at local materials instead if you would rather not add one.
- Protocol
- REST / OpenAPI
- Endpoint
https://moneyprinterturbo.example.com- Authentication
- bearer
Where to add it
Open Dashboard Tools, select a running OpenClaw or Hermes instance, then use the card’s Install or Connect action. External services may require an HTTPS endpoint and credentials; the dashboard shows those fields before anything is saved.