Skip to content
Image Asset Server

Image set builder

Every image a page needs, licence-checked, in one go

Write the images you need: a hero, a few icons, a background, an illustration. Each one is searched across every source with its shape, size and kind, and the best image with direct files is picked, with alternatives. The images stay on the sources: the exports point at their files.

One image per line, described like a search (“vertical”, “png”, “icon”, “4k” work too). End a line with “(1200px)” to set its width; “3 icons: rocket, shield, chart” makes one slot per item. Up to 24.

Orientation
Kind

How it picks the images

  1. 1List the images. One image per line, described like a search ('hero: mountain lake at sunrise'). Add '(1200px)' to set a slot's width; '3 icons: rocket, shield, chart' makes one slot per item.
  2. 2Set the defaults. A width, an orientation and a kind (photo, illustration, painting, vector, icon) for slots that don't say. Commercial-safe is on by default.
  3. 3Review the picks. Each slot gets the best matching image with direct files, at least half the slot's width where possible, plus alternatives. Check the verdicts and notes.
  4. 4Export. Download images.json for your code, credits.html for <figure> snippets with credit captions, download.sh to fetch the files, and ATTRIBUTION.md for the credits.

An image is never used twice. For each pick the smallest file at least as wide as the slot is chosen (a 1920 px copy, not a 60 MB TIFF). When a pick is smaller than its slot, the wrong shape or editorial-only, the slot says so.

What the exports do

images.json
Every pick with its file URL, pixel size, format, suggested alt text, licence, verdict and credit line, ready for a static site or a slide generator.
credits.html
One <figure> per image with width, height, alt text and the credit in the <figcaption>: what CC BY asks for.
download.sh
A POSIX shell script that curls every pick into ./images with tidy file names, straight from the sources.
ATTRIBUTION.md, credits.json
Every image with its source, author, licence and credit line, for a credits page or your repository.

Guide: let an AI agent find and license images →

From code or an agent

The same set over the REST API:

curl -X POST https://image.shep.bot/v1/sets \
  -H "content-type: application/json" \
  -d '{"slots":["hero: mountain lake at sunrise (1920px)","3 icons: rocket, shield, chart","botanical illustration of a rose"],"width":1600}'

From a terminal: npx image-asset-server set slots.txt --width 1600 --download. Or connect https://image.shep.bot/mcp to your agent (as images) and ask for “images for a landing page: a hero, three feature icons and a background”. Run locally, build_image_set with download=true also saves the images next to the exports, which then use the local files. Agent setup