@DanKornas: Websites often have existing APIs, forms, and authenticated state that AI agents cannot call as structured tools from t…

X AI KOLs Timeline Tools

Summary

WebMCP is a protocol implemented by MCP-B that lets websites act as MCP servers, exposing browser JavaScript functionality to LLMs as tools via tab or extension transports. The codebase is maintained for historical reference only.

Websites often have existing APIs, forms, and authenticated state that AI agents cannot call as structured tools from the browser. WebMCP is the protocol implemented by MCP-B, an older historical-reference codebase that lets a website act as an MCP server by exposing browser JavaScript functionality to LLMs as MCP tools. A website creates an MCP server, wraps its client-side logic as tools, and connects through tab or extension transports so an AI client can discover and invoke those tools in the browser context. Key features: • Tab Transports use postMessage for communication between a website MCP server and clients in the same tab. • Extension Transports use Chrome runtime messaging to connect browser extensions to website tools. • Tools can wrap existing APIs, forms, or state and remain scoped to specific pages or user states. • The native server proxies requests from local MCP clients to the browser extension so desktop applications can call tools from open website tabs. • Tools run in the page context, respect the browser sandbox and same-origin policy, and can be audited through the extension. The README lists the license as AGPL-3.0, while also stating that the MCP-B extension is no longer open source and this repository is maintained for historical reference only. Link in the reply
Original Article
View Cached Full Text

Cached at: 08/03/26, 05:47 PM

Websites often have existing APIs, forms, and authenticated state that AI agents cannot call as structured tools from the browser.

WebMCP is the protocol implemented by MCP-B, an older historical-reference codebase that lets a website act as an MCP server by exposing browser JavaScript functionality to LLMs as MCP tools.

A website creates an MCP server, wraps its client-side logic as tools, and connects through tab or extension transports so an AI client can discover and invoke those tools in the browser context.

Key features: • Tab Transports use postMessage for communication between a website MCP server and clients in the same tab. • Extension Transports use Chrome runtime messaging to connect browser extensions to website tools. • Tools can wrap existing APIs, forms, or state and remain scoped to specific pages or user states. • The native server proxies requests from local MCP clients to the browser extension so desktop applications can call tools from open website tabs. • Tools run in the page context, respect the browser sandbox and same-origin policy, and can be audited through the extension.

The README lists the license as AGPL-3.0, while also stating that the MCP-B extension is no longer open source and this repository is maintained for historical reference only.

Link in the reply

Similar Articles

Introducing the MDN MCP server

Lobsters Hottest

Mozilla announced the MDN MCP server, which uses the Model Context Protocol to provide AI coding tools with up-to-date MDN documentation and browser compatibility data, enabling more accurate web platform information in development workflows.

API to MCP

Product Hunt

API to MCP lets you turn any API into an MCP server for AI agents, enabling seamless integration.