What it does
The server scans a public web page and reports on its images: natural versus rendered dimensions, images that are oversized for the space they render into, and a breakdown by format. The result is a text report with structured content. It includes total and visible image counts, the number of CSS background images, the share of modern formats (WebP and AVIF), a format breakdown, and a list of oversized images showing natural and rendered size, overshoot ratio, format and source. It also links to a full visual report on imagedimensions.com.
AI agents such as Claude, Cursor and Windsurf can call it during development to check image performance. Typical uses are finding oversized images that may slow Largest Contentful Paint, and measuring what share of a site's images use modern formats. An image downloaded far larger than its rendered box uses extra bandwidth and slows page loading.
The scan runs on the server side, so no local browser or Chrome installation is needed. The required input is the page URL. An optional oversizedThreshold sets the area-overshoot ratio used to flag oversized images; the default is 4, roughly 2x per dimension. The API base can be changed with the IMAGEDIMENSIONS_API_BASE setting. The project is released under the MIT license.
Limitations and requirements
Only public web pages can be audited. The scan is performed by the imagedimensions.com service, and the default configuration points to it.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Official MCP Registry first seen 13 Nov 2026 · last seen 13 Nov 2026 · its page on the source
io.github.Bishop81/imagedimensions-mcp
Audit a web page's images: natural vs rendered sizes, oversized detection, and formats.
v0.1.2As shown on registry.modelcontextprotocol.io when we last read it