Website Image Downloader — API & MCP examples

Name: Website Image Downloader
Author: sstempresarial

Download all images from any website (or just extract their URLs) with width, height, format, file size and alt text — as a ZIP, through a REST API, or as a tool for AI agents via MCP. Icons, sprites and tracking pixels are filtered out; srcset, <picture> and lazy-loaded images are handled.

This repo contains ready-to-run examples for the hosted Website Image Downloader (Apify Actor sste/website-image-downloader), plus dependency-light do-it-yourself extractors in Python and JavaScript.


Store page	https://apify.com/sste/website-image-downloader
Docs & guides	https://website-image-downloader.pages.dev
MCP server	`https://mcp.apify.com?tools=sste/website-image-downloader`
Pricing	$0.20 per 1,000 images · $0.05 per 1,000 URLs in list mode · $0.001 per run

Quick start (cURL)

bash

export APIFY_TOKEN=...   # Apify Console → Settings → Integrations
bash curl/download-all-images.sh https://www.example.com/
bash curl/list-image-urls.sh https://www.example.com/

Python

bash

pip install "apify-client>=3"
python python/download_images.py https://www.example.com/     # writes images.zip + images.csv
python python/diy_extract_image_urls.py https://www.example.com/  # no API, requests + BeautifulSoup

JavaScript / Node.js

bash

npm i apify-client cheerio
node javascript/download-images.mjs https://www.example.com/
node javascript/diy-extract-image-urls.mjs https://www.example.com/

AI agents (MCP)

Client	Config
Claude Desktop / Cursor / Windsurf	`mcp/claude_desktop_config.json`
VS Code (`.vscode/mcp.json`)	`mcp/vscode-mcp.json`
Claude Code	`claude mcp add --transport http website-image-downloader "https://mcp.apify.com?tools=sste/website-image-downloader"`
Any client (stdio)	`npx -y @apify/actors-mcp-server --tools sste/website-image-downloader`

Example prompts: mcp/prompts.md. Registry entry: server.json (io.github.SSTEmpresarial/website-image-downloader).

What the hosted tool handles

largest srcset / <picture> candidate, lazy-load attributes (data-src, data-srcset, data-lazy-src, data-original, data-bg), <noscript>, CSS backgrounds, og:image
real width/height/format from file bytes; filters by size, format and URL text; soft-404 detection; SHA-256 dedupe
modes: download (ZIP + rows), inspect (rows only), list (URLs only)
same-site crawling (depth 1–3), spending caps