generate_and_wait joins submission and polling together, so a paragraph of prose comes back as a finished visual without the client tracking request IDs. The separate submit-and-poll path stays available for a batch that would otherwise block for too long. list_styles reports the styles on offer but points at the Napkin styles documentation for the full list, so a style_id may still need looking up there.
A Napkin AI client that submits visual generation requests and collects the files they produce.
- Visual generation from content plus context, with control over format, orientation, width and height, colour mode, transparent background, style ID, language, how many visuals to make and how they are sorted.
- A choice of loop: submit and poll the status yourself, or make the single call that waits for completion.
- Delivery either as base64 data pulled from the file URL a status response carries, or written straight into storage configured on the server.
- A dry run, a style listing and an API key check, for confirming a setup before a real generation.
NAPKIN_API_KEY, which verify_api_key exists to confirm. generate_and_save additionally needs storage configured in the server settings; without it, finished files come back through download_visual.
One command plus a key — npm install -g napkin-ai-mcp, then supply credentials