README.md

metadata

title: Prompt Groomer Demo
emoji: 🧹
colorFrom: blue
colorTo: green
sdk: streamlit
sdk_version: 1.28.0
app_file: app.py
pinned: false

🧹 Prompt Groomer - Interactive Demo

Live demonstration of prompt-groomer library capabilities.

Stop paying for invisible tokens. Optimize your LLM inputs to save costs, improve context usage, and enhance security.

✨ Features

• 🎯 Interactive Playground - Experiment with different optimization strategies
• 💰 Real-time Token Savings - See exactly how many tokens and dollars you save
• 🔧 Configurable Operations - Toggle between 7 different operations
• 📊 Visual Metrics - Cost analysis and performance tracking
• 🎭 Preset Examples - 6 real-world scenarios (e-commerce, support, docs, RAG, etc.)
• ⚡ Quick Presets - Minimal, Standard, Aggressive, or Custom strategies

🚀 Quick Start

1. Choose a preset example from the dropdown (or enter your own text)
2. Select a strategy in the sidebar (Minimal, Standard, Aggressive, or Custom)
3. Configure operations to match your use case
4. See results with real-time token counting and cost savings

📦 Installation

Try it live here, then install for your project:

pip install prompt-groomer

💻 Example Usage

from prompt_groomer import (
    StripHTML,
    NormalizeWhitespace,
    TruncateTokens
)

# Use pipe operator to chain operations
pipeline = (
    StripHTML()
    | NormalizeWhitespace()
    | TruncateTokens(max_tokens=1000)
)

cleaned = pipeline.run(dirty_text)

🔗 Links

• 📖 Documentation
• 💻 GitHub Repository
• 📦 PyPI Package

📊 Proven Effectiveness

Benchmarked on 30 real-world test cases:

• 4-15% token reduction on average
• 96-99% quality maintained (verified)
• Up to ~$54/month saved at scale (1M tokens/month, GPT-4)

🛠️ Operations Available

🧼 Cleaner

• Strip HTML tags
• Normalize whitespace
• Fix Unicode issues

🗜️ Compressor

• Deduplicate similar content
• Truncate to token limits

🔒 Scrubber

• Redact PII (email, phone, IP, credit cards, SSN, URLs)

📊 Analyzer

• Count tokens
• Calculate cost savings

---

Made with ❤️ by Jacob Huang