Google AI Overview Scraper 作者： BalaK

Scrape Google AI Overviews into clean, structured Markdown — no API keys, no accounts, no data leaving your machine.

Google AI Overview Scraper is a Firefox extension that acts as a local relay between Google Search and a small MCP (Model Context Protocol) server you run yourself. AI coding tools — Claude Code, Claude Desktop, Cursor, and any MCP-compatible client — can search Google AI Overviews and get back well-formed Markdown with source citations, all through your own browser.

No focus stealing. Queries run in background tabs, so the extension never interrupts what you're doing.

How it works
1. Your AI tool sends a search query through the local MCP server.
2. The extension opens a background tab on Google Search (no focus stealing).
3. The AI Overview is extracted from the page and converted to structured Markdown.
4. Results are sent back to your AI tool together with their source citations.

Features
- No API keys or accounts — it uses your own Firefox session.
- Background tabs — never steals focus from your work.
- Structured Markdown output with source citations.
- Conversational follow-ups via thread IDs (threads auto-expire when idle).
- Configurable server URL from the options page.
- Three simple MCP tools: search, follow_up, and health.
- Works with Claude Code, Claude Desktop, Cursor, and any MCP client.

Privacy first
Everything stays on your machine. The extension talks only to a local server on localhost that you run and control — there is no analytics, no tracking, and no third-party data collection. It runs only on Google Search result pages, and the add-on declares "no" data collection to Firefox.

Who it's for
Developers and power users who want to give an AI assistant clean, citation-backed access to Google AI Overviews — for research, summarization, and automation — without sending anything to a third party.

Setup
Install the companion MCP server and point the extension's options page at it:

uvx google-ai-scraper

Then set the server URL on the options page to http://localhost:15551 and add the server to your MCP client:

{
"mcpServers": {
"google-ai-scraper": {
"command": "uvx",
"args": ["google-ai-scraper"]
}
}
}

Full instructions, the HTTP API, and source code:
https://github.com/balakumardev/google-ai-scraper