Browser Extensions & ToolsStatus: completedAugust 15, 2026
Gem Harvester — Autonomous Web Scraping & Workflow Automation Suite
Intelligent web scraping & workflow automation suite built on Plasmo, React, Hugging Face Transformers, and Ant Design for AI data harvesting.
Tech Stack:PlasmoReactTypeScriptTransformersDexie.jsAnt DesignDOM AutomationChrome Extension
Direct Suite Extension Package & Installation
- ⚡ Download Gem Harvester Suite (.zip) — Pre-built production Chromium package (~1.95 MB).
- 💻 GitHub Repository — Extension source code and routine engine.
How to install in Chrome / Brave / Edge:
- Download and extract the
.ziparchive to a folder. - In your browser, open
chrome://extensions/. - Enable Developer mode in the upper right corner.
- Click Load unpacked and select the extracted folder.
Overview
Gem Harvester (v0.9.1) is a specialized browser automation and intelligent data extraction suite engineered for complex web harvesting workflows. Built with Plasmo, React, Ant Design, and in-browser AI models via @huggingface/transformers, the suite empowers researchers and developers to construct multi-step visual scraping routines, extract structured schemas, clean data on-device with zero API fees, and store records in Dexie (IndexedDB).
Key Architecture & Capabilities
- 🔄 Visual Routine Pipeline Builder:
- Construct automated workflows consisting of Navigation, Delay/Wait, Click, Form Input, DOM Extraction, Infinite Scroll, and Conditional Loops.
- 🤖 On-Device AI Entity Extraction:
- Leverages client-side
@huggingface/transformersto classify, extract, and clean unstructured DOM text directly in the browser tab.
- Leverages client-side
- 📊 Adaptive DOM Extraction Engine:
- Heuristic pattern detection across HTML tables, dynamic e-commerce card grids, and paginated catalogs with CSS selector overrides.
- 💾 High-Capacity IndexedDB Persistence:
- Built-in Dexie.js database engine storing harvested datasets locally without localStorage quota limits.
- ⚡ Instant Multi-Format Export:
- One-click dataset export to formatted JSON, clean CSV (via PapaParse), or clipboard.
- 🎨 Polished Ant Design Cockpit:
- Native drawer runner with live step-by-step telemetry, execution progress bars, error handling, and dark/light mode sync.
Tech Stack
- Extension Framework: Plasmo (Manifest V3)
- UI Framework: React 18, TypeScript, Ant Design, Tailwind CSS
- AI & Data Engine:
@huggingface/transformers, Dexie.js (IndexedDB), PapaParse