Make any website agent-ready in one line

WebMCP exposes a site's tools to in-browser AI agents through navigator.modelContext. Drop in one script tag and your pages become agent-callable. The same extraction also emits a server-side MCP endpoint, so you cover both protocols from one source.

🔌 The MCP hub 🛡️ Trust leaderboard 📚 Directory

Generate your shim

Your product / page URL

Paste this one line into your page <head>:

<script src="https://wmcp.sh/webmcp/aHR0cHM6Ly93d3cuYWxsYmlyZHMuY29tL3Byb2R1Y3RzL21lbnMtd29vbC1ydW5uZXJz.js" defer></script>

Read tools resolve instantly from the embedded extraction. Live actions route through wmcp.sh's graded, metered execute path.

How it works

1 · One extraction

wmcp.sh reads your page (JSON-LD, OpenAPI, Shopify, and more) and turns it into tools.

2 · Dual-emit

The same tools serve as a WebMCP shim for in-browser agents and a server-side MCP endpoint for Claude, Cursor, and Codex.

3 · Governed actions

Reads are free and instant. Live actions flow through the metered, graded execute path with kill switch and audit.

Why WebMCP

Browser agents are arriving, and they look for tools on the page via navigator.modelContext. Sites that expose tools get used by those agents; sites that don't are invisible to them. This is the cheapest way to be on the right side of that shift, and it doubles as a server-side MCP endpoint.

Grade, connect and build at the MCP hub →