LionScraper MCP + CLI + HTTP API bridge

简体中文

• Website: lionscraper.com
• npm: package lionscraper
• PyPI: project lionscraper

What is this?

LionScraper is a browser extension that can collect lists, articles, links, images, and more from web pages. This repository provides the companion bridge between your tools and that extension in three ways:

• MCP (lionscraper-mcp): connect an AI app (e.g. Cursor) so the model can call scraping tools over stdio.
• CLI (lionscraper): run daemon, scrape, ping, and more from a terminal on the same local HTTP/WebSocket port as the extension.
• HTTP API: when the daemon is running, call the same capabilities over loopback JSON HTTP (e.g. /v1/...) from scripts or any HTTP client—no MCP or CLI front-end required.

The real scraping logic runs in the extension; these packages connect and forward.

Before you start

1. Browser: Chrome or Edge (follow what the extension supports).
2. LionScraper extension: install and enable from the store.
   • Chrome: Chrome Web Store — LionScraper
   • Microsoft Edge: Edge Add-ons — LionScraper
3. Runtime (pick one or both implementations):
   • Node.js 18+ for the npm package — Node.js
   • Python 3.10+ for the PyPI package — Python
4. For MCP: an AI app that supports MCP (e.g. Cursor, Trae).
5. For the HTTP API: same browser, extension, and daemon as the CLI; see the package READMEs for paths and examples.

HTTP fallback without Chrome/Edge: If neither browser is detected under standard paths and the extension is not connected, MCP still starts; ping succeeds with http_fetch mode and scrape* use a minimal server-side HTTP GET (no JS execution). If a browser is installed but the extension is not connected, you still get the extension connection flow. The Node auto-spawn path fixes Unix installs where lionscraper.js was resolved without a leading / (e.g. Glama/Docker). The Python package uses aiohttp for outbound HTTP/WebSocket to the daemon.

Two implementations

| | Node.js (npm) | Python (pip) | |--|-------------------|------------------| | Registry | io.github.dowant/lionscraper-node | io.github.dowant/lionscraper-python | | Docs (EN) | packages/node/README.md | packages/python/README.md | | Docs (ZH) | packages/node/README_cn.md | packages/python/README_cn.md |

Install one or both; they are separate packages with the same CLI command names.

Install (npm)

Published as lionscraper on npm.

npm install -g lionscraper

Without a global install, MCP can use npx; see the npx JSON examples under Add MCP in your AI app.

Install (pip)

Published as lionscraper on PyPI.

pip install -U lionscraper

A virtual environment is recommended, or pip install -U --user lionscraper if you prefer not to install into the system interpreter.

Commands (both packages)

| Command | Role | |--------|------| | lionscraper-mcp | Thin MCP server (stdio) for AI apps | | lionscraper | CLI: daemon, stop, scrape, ping, … (also serves the HTTP API on the same port) |

After pip install -U lionscraper, if lionscraper-mcp is not on your PATH, use python -m lionscraper with no extra arguments for MCP stdio (see [packages/python/README.md](packa

... View full README on GitHub