ragzip

RAGZip

Built with Hyperbrowser

A single-file CLI that uses Hyperbrowser's official scrape API to extract website content and builds citation-tagged context packs for LLMs.

Setup

1. Get an API key from https://hyperbrowser.ai

2. Install dependencies

   npm install

3. Set environment variables

   cp env.example .env
   # Add your keys to .env:
   # HYPERBROWSER_API_KEY=hb_your_key_here
   # OPENAI_API_KEY=sk_your_key_here  # optional for compression

Usage

# Basic usage
npm start -- --url https://example.com --budget 2000

# Multiple URLs with compression
npm start -- --url https://docs.example.com --url https://blog.example.com --llm --format md

# From file
echo "https://example.com" > urls.txt
npm start -- --urls urls.txt --budget 5000

Options

• --url <url> - URLs to process (can repeat)
• --urls <file> - File with newline-separated URLs
• --budget <n> - Token budget (default: 8000)
• --out <dir> - Output directory (default: "distill")
• --format <jsonl|md> - Output format (default: jsonl)
• --llm - Enable OpenAI compression

Output

• pack.jsonl - Main context pack with chunks
• stats.json - Processing statistics
• citations.md - Source URLs

Follow @hyperbrowser_ai for updates.