What is WebMCP, and why does it matter?
What is WebMCP?
WebMCP (Web Model Context Protocol) lets websites expose tools that in-browser AI agents can call. Instead of scraping HTML or guessing which buttons to click, an agent can use document.modelContext to discover and run structured tools your page registers — crop an image, resize it, convert a format, or fetch page context. Think of llms.txt as a map of your content for language models, and WebMCP as a remote control for interactive features that live in the browser.
How is it different from llms.txt?
llms.txt and llms-full.txt are static Markdown files. They help models read and cite your documentation. WebMCP is runtime and interactive. Tools register after a page loads, usually via document.modelContext. Agents open your page in a browser tab, then call tools with typed inputs (for example set_crop_aspect_ratio or download_result). There is typically no headless server API — processing happens client-side.
Why publish a webmcp.json file?
A discovery file at /.well-known/webmcp.json (or a similar well-known path) describes your site's WebMCP tools: names, schemas, which pages they belong to, and limitations. Agents and directories can find that file the same way they look for robots.txt or llms.txt — before they open every route. Publishing it makes your site easier to adopt for browser AI agents and clearer for developers exploring WebMCP.
What does a typical setup look like?
1. Host a JSON discovery document (often at /.well-known/webmcp.json). 2. List site-level tools and page-specific tools with input schemas. 3. On each relevant page, register those tools with document.modelContext after load. 4. Document limitations: CORS, user confirmation for downloads, and the need for an open browser tab. Chrome's WebMCP documentation and the Web Machine Learning WebMCP draft are good starting points for the API surface.
Why list it on llmslist?
llmslist already indexes sites that publish llms.txt. Many of those same products are adding WebMCP so agents can act, not only read. When you submit your site, you can optionally add your webmcp.json URL. Visitors and agents then see both your content map and your interactive tool surface in one place.
Useful links
FAQ
What is WebMCP?
WebMCP is a way for websites to register browser-side tools that AI agents can call through document.modelContext, so agents can use your UI features in a structured way.
Where should webmcp.json live?
Common practice is to publish a discovery file at /.well-known/webmcp.json (or a documented well-known URL) that lists your tools, pages, and schemas.
Do agents need a server API?
Usually no. WebMCP tools often run in the open browser tab after the page loads. Local file pickers and downloads may still require user confirmation, and image fetches can be limited by CORS.
How does WebMCP relate to llms.txt?
llms.txt helps models understand your content; WebMCP helps agents operate your interactive tools. Many sites will want both.
Have a webmcp.json?
Submit your site to llmslist and include your optional WebMCP discovery URL so agents can find your tools.
Submit Your Site