promptarium

api & mcp

Use promptarium from Claude, Cursor or your own code. Searching the prompt library is free. Generating uses your credits, at the same prices as the studio.

get your API key

Connect an AI app (MCP)

The server address is https://promptarium.app/mcp. Add it to any app that takes MCP servers over HTTP, with your key in the Authorization header. Without a key the library tools still work.

Claude Code

claude mcp add --transport http promptarium https://promptarium.app/mcp --header "Authorization: Bearer YOUR_KEY"

Cursor and other apps

{
  "mcpServers": {
    "promptarium": {
      "url": "https://promptarium.app/mcp",
      "headers": { "Authorization": "Bearer YOUR_KEY" }
    }
  }
}

Then ask for what you want, for example: "find the most viewed Seedance UGC prompts and make the best one into a 5 second video".

toolwhat it doeskey
search_promptsSearch the prompt library by words, model and type. Sort by relevance, views or newest.no
get_promptThe full text of one prompt, with its result, creator and source post.no
list_modelsEvery image and video model, its options and its price in credits.no
estimate_costWhat a generation would cost with exact settings. Makes nothing.no
get_plansPlans and top-ups, with the links to buy them.no
generate_imageMakes 1 to 4 images and waits for them.yes
generate_videoStarts a video and returns its id.yes
get_generationStatus of a generation and the link to its file. Can wait for it.yes
list_generationsThe account's latest images and videos.yes
get_balanceCredits left and the current plan.yes

REST API

Base address: https://promptarium.app/api/v1. Send your key as Authorization: Bearer YOUR_KEY. Answers are JSON. An error has an error message and a matching status code.

Search the prompt library

curl "https://promptarium.app/api/v1/prompts?q=ugc+skincare&type=video&limit=5"

Make an image

curl https://promptarium.app/api/v1/images \
  -H "Authorization: Bearer YOUR_KEY" -H "Content-Type: application/json" \
  -d '{"prompt": "a red fox in fresh snow, golden hour", "model": "Nano Banana 2", "aspect_ratio": "16:9", "wait": 45}'

Fields: prompt, model, aspect_ratio, resolution, quality, count (1 to 4), reference_images (https links, or ids of images you made here), max_credits, wait (seconds, up to 55).

Make a video

curl https://promptarium.app/api/v1/videos \
  -H "Authorization: Bearer YOUR_KEY" -H "Content-Type: application/json" \
  -d '{"prompt": "handheld selfie vlog on a rainy street, she laughs", "model": "Seedance 2.5", "duration": 5, "max_credits": 200}'

curl "https://promptarium.app/api/v1/generations/GENERATION_ID?wait=50" -H "Authorization: Bearer YOUR_KEY"

Fields: prompt, model, duration, aspect_ratio, resolution, audio, first_frame, last_frame, max_credits. Videos take 1 to 10 minutes, so ask for the generation until its status is done and it has a url.

All endpoints

endpointwhat it doeskey
GET /promptsSearch the library. q, model, type, sort, limit, offset.no
GET /prompts/{id}One prompt in full. The id looks like seedance-2/0001.no
GET /librariesHow many prompts each model's library holds.no
GET /modelsEvery model with its options and price. type=image or type=video.no
POST /estimateThe price of a generation with exact settings.no
GET /plansPlans and top-ups.no
GET /balanceCredits and plan.yes
POST /imagesMake 1 to 4 images.yes
POST /videosStart a video.yes
GET /generationsYour latest generations.yes
GET /generations/{id}One generation. wait=50 waits up to 50 seconds for it.yes

Prices and limits

Keys

The library's prompts

Every library prompt comes with its creator and a link to the original post. If you show a prompt somewhere else, keep the credit and link to its promptarium page.