webmcp-radar

First public registry of WebMCP-enabled websites. Continuously scans candidate domains for the /.well-known/webmcp discovery endpoint and tracks adoption over time.

last scan: —
—
live sites
—
tools indexed
—
new last 24h
—
dead / pending
loading…
loading…

We will fetch /.well-known/webmcp, then /.well-known/mcp, then look for WebMCP script tags on the home page. Results are public and appear on the leaderboard. Rate-limited to 10 submissions per minute per IP.

What is this?

WebMCP entered a public origin trial in Chrome 149 on May 19, 2026 (Microsoft shipped Edge 147 support back in March). It lets websites expose structured tools to in-browser AI agents via the W3C Web Machine Learning Community Group spec. As of launch there was no public registry of WebMCP-enabled sites — the spec explicitly flags discovery as unsolved. webmcp-radar fills that gap.

How discovery works

  1. Fetch https://<domain>/.well-known/webmcp
  2. If 404, fall back to /.well-known/mcp (SEP-1960)
  3. If still nothing, scrape the home page for WebMCP script / meta tags

Where the candidate domains come from

  • Curated seed list (~80 well-known dev/AI sites), every 6h
  • Hacker News top 50 stories, every 2h
  • GitHub trending repo READMEs, every 12h
  • User submissions via the form above

Honesty contract

Every site shown here was actually fetched by us. Every tool count comes from a real WebMCP manifest. No seeded fake sites, no mocked tools. If the list looks empty, that is the truth of WebMCP adoption right now.