Package Exports

@gecho-ai/bridge
@gecho-ai/bridge/mcp-client.js

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@gecho-ai/bridge) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

Gecho Bridge 🚀

Gecho Bridge is an MCP (Model Context Protocol) server that connects AI Agents (like Claude Desktop, Cursor, or OpenClaw) to TikTok via a Chrome extension. It allows your AI models to search and retrieve TikTok video metadata directly from the browser.

Features

Automated Search: Search TikTok with keywords and auto-scroll to fetch results.
Data Retrieval: Get video IDs, titles, like counts, and play URLs.
Dual-Layer Architecture: A lightweight MCP Client that automatically manages a persistent Service Layer.
Lazy Start: The bridge automatically starts the background service when needed.

Installation

1. Install via npm

npm install -g gecho-bridge

2. Install Chrome Extension

Ensure you have the Gecho TikTok Extension installed and active in your Chrome browser. The bridge communicates with this extension via WebSocket.

Configuration

Claude Desktop

Add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "gecho-bridge": {
      "command": "npx",
      "args": ["-y", "gecho-bridge"]
    }
  }
}

Cursor

Go to Settings -> MCP.
Add a new MCP server:
- Name: gecho-bridge
- Type: command
- Command: npx -y gecho-bridge

Usage

Once configured, you can ask your AI:

"Search TikTok for 'cooking recipes'"
"Find trending TikTok videos about AI agents"

Development

If you want to run the bridge locally from source:

git clone https://github.com/gecho-ai/bridge.git
cd bridge
npm install
node mcp-client.js

Architecture

Service Layer (server.js): Maintains the WebSocket connection with the Chrome extension (Port 18792) and exposes an HTTP API (Port 18793).
Client Layer (mcp-client.js): Implements the MCP Stdio protocol and forwards requests to the Service Layer.

License

MIT