Developer API & open data

Every prompt in the Promptkoi library — 210 of them, each tested on the model it targets and published with its exact settings — is available as free, static JSON. No API key, no rate limit, no sign-up. Attribution (a link back to promptkoi.com) is appreciated and, for the dataset, requested.

The full dataset

GET /data/prompts.json returns the entire library: title, prompt text, negative prompt, model + version, settings, tags, tested date, and canonical URL for every entry. It rebuilds on every deploy, so it's always current.

A single prompt

Any prompt page has a “Copy as JSON” block (under “For developers”) returning just that entry. The shape matches one element of the dataset array:

{
  "title": "Neon Cyberpunk City Street",
  "prompt": "neon-drenched cyberpunk city street at night, … --ar 16:9 --stylize 250",
  "model": "Midjourney",
  "model_version": "v7",
  "settings": {
    "aspect_ratio": "16:9",
    "stylize": 250
  },
  "style_tags": [
    "cyberpunk",
    "neon"
  ],
  "use_case_tags": [
    "wallpaper"
  ],
  "url": "https://promptkoi.com/prompts/midjourney/neon-cyberpunk-city/"
}

Other machine-readable endpoints

• /rss.xml — full site feed (WebSub-enabled); per-model feeds at /{model}-prompts/rss.xml.
• /sitemap-index.xml and /image-sitemap.xml — every URL with truthful lastmod.
• /llms.txt — a curated index for AI answer engines.
• /ai-image-statistics/ — aggregate corpus statistics, free to cite with attribution.

License & attribution

The prompts are free to use to generate images. The dataset text is © Promptkoi; if you republish or build on it, please credit and link promptkoi.com. See the terms for details.