Home › Guides

Website images for AI agents: download or list images via MCP

LLM agents can browse text, but images on a page are hard: they need the real file URLs (not thumbnails), sizes to choose the right one, and a way to hand files to the user. An MCP tool solves this.

Setup (one URL)

https://mcp.apify.com?tools=sste/website-image-downloader

Client-specific configs (Claude Desktop, Claude Code, Cursor, VS Code): MCP setup.

What the agent can do

• List images with width, height, format, size and alt text (mode: "inspect" or "list").
• Download them and return the ZIP link (mode: "download").
• Audit pages: missing alt text, oversized files, legacy formats.

Example prompts

• "Get all images over 800 px from https://example.com/press and summarize what they show based on alt text."
• "Download the product photos from these 10 product URLs as a ZIP."
• "Which images on our homepage have no alt text?"

Related

• How to download all images from a website (and get them as a ZIP)
• How to extract image URLs from a webpage (srcset, lazy-loaded, CSS)
• How to filter icons and tracking pixels when scraping images
• Scraping srcset and lazy-loaded images correctly