BrowserBee MCP Demo Server

Overview

The BrowserBee MCP Demo is a lightweight reference implementation that demonstrates how an MCP (Model Context Protocol) server can be embedded directly into a web application and made discoverable by the BrowserBee browser extension. It solves a common pain point for developers building AI‑enabled web experiences: the need to expose structured, machine‑readable context and tools from a browser environment without building a full backend. By running the MCP server in the client’s browser, developers can let AI assistants query real‑time page data, invoke JavaScript functions, or interact with the browser’s DOM—all while keeping the entire workflow within a single web page.

At its core, the demo adds three key capabilities to standard browser messaging. First, every message is prefixed with so the BrowserBee extension can filter and route MCP traffic. Second, it injects and metadata into each message, allowing the server to maintain session context and identify its origin. Finally, a simple message advertises the MCP server’s presence to BrowserBee, enabling automatic discovery and connection without manual configuration. These enhancements are sufficient for a minimal MCP server to operate in the browser, yet they illustrate the protocol’s extensibility.

Key features of this demo include:

• In‑browser MCP server: Runs entirely in the page, eliminating the need for a separate backend service.
• Seamless BrowserBee integration: Uses to communicate with the extension, ensuring low latency and a native feel.
• Session awareness: The added metadata lets the server track conversational state across multiple messages.
• Extensibility hooks: Developers can replace or extend the to expose additional tools, such as scraping APIs or custom JavaScript functions.

Typical use cases for this server are:

• AI‑powered web assistants that need to read or modify page content in real time.
• Testing and prototyping of MCP‑compatible agents without deploying a full server stack.
• Educational demos that show how browser extensions and web apps can cooperate under the MCP framework.

By embedding an MCP server in a web page, developers can quickly prototype AI workflows that involve dynamic data from the browser. The demo showcases how a minimal set of protocol conventions—message prefixes, metadata fields, and a ping advertisement—are enough to turn any web app into an AI‑ready endpoint. This approach gives teams a fast, low‑overhead path to experiment with advanced AI interactions directly in the browser.