redukon 3.2.0

Simple CLI tool for AI text rewriting with Ollama

🔥 Redukon

Reduce Your Tokens. Keep Your Intent.

---

✨ What is Redukon?

Redukon is a token-saving prompt rewriter that uses local small AI models to optimize your prompts — reducing token count while keeping the core intent intact.

No API keys needed. Runs entirely offline with Ollama.

💡 Perfect for:

• Saving money on API calls
• Fitting more context into LLM windows
• Making prompts more efficient
• Building into your own apps via API

---

🚀 Quick Start

# Install from PyPI
pip install redukon

# Or install latest from GitHub
pip install -e https://github.com/CharlesArea/Redukon.git

# One-time setup (installs Ollama + picks a model)
redukon onboard

---

📖 CLI Usage

redukon onboard

Interactive setup wizard:

1. Checks if Ollama is installed
2. Installs Ollama if needed
3. Choose your model:

   Model	Size	Best For
   qwen2.5:0.5b	397MB	Speed & minimal resources
   llama3.2:1b	1.9GB	Balanced
   phi3:3.8b	2.3GB	Better quality

redukon rewrite

Rewrite your prompts from command line:

# Basic
redukon rewrite -i "Please write a comprehensive Python function..."

# From file
redukon rewrite -i @long-prompt.txt -o optimized.txt

# Custom temperature (lower = more focused)
redukon rewrite -i "prompt" --temp 0.3

redukon serve

Start the API server:

# Default port 8000
redukon serve

# Custom port
redukon serve --port 9000
redukon serve --host 127.0.0.1 --port 8080

---

🌐 API Usage

Start Server

redukon serve

Endpoints

Endpoint	Method	Description
/rewrite	POST	Rewrite a prompt
/health	GET	Health check

Example Request

curl -X POST http://localhost:8000/rewrite \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Please write a comprehensive Python function that takes a list of integers...",
    "model": "qwen2.5:0.5b",
    "temperature": 0.3
  }'

Example Response

{
  "optimized_prompt": "Write a Python function that filters even numbers from a list...",
  "original_tokens": 87,
  "optimized_tokens": 42,
  "saved_tokens": 45,
  "saved_percent": 52
}

API Parameters

Parameter	Type	Required	Description
prompt	string	Yes	The prompt to optimize
model	string	No	Model name (default: from config)
temperature	float	No	Temperature 0.0-1.0 (default: 0.3)

---

📝 Logging

API requests are logged to log/api-YYYY-MM-DD.log:

[2026-03-10 12:00:00] [REQUEST] input_length=250, model=qwen2.5:0.5b
[2026-03-10 12:00:05] [RESPONSE] output_length=120, original_tokens=62, optimized_tokens=30, saved_tokens=32, saved_percent=51%
[2026-03-10 12:00:05] [ERROR] Generation failed: Ollama not running

---

⚡ Example

Before (87 tokens)	After (42 tokens)
"Please write a comprehensive Python function that takes a list of integers as input and returns a new list containing only the even numbers from the original list. The function should handle edge cases like empty lists, lists with no even numbers, and lists with negative numbers. Please include proper type hints, docstrings, and error handling."	"Write a Python function that filters even numbers from a list. Include type hints, error handling, and docstrings."

📉 Savings: ~52%

---

🔧 Options

Flag	Description	Default
-i, --input	Prompt or @file.txt	Required
-o, --output	Output file	Print to stdout
-m, --model	Override model	From config
-t, --temp	Temperature (0.0-1.0)	0.3

---

🛠️ Requirements

• Python 3.10+
• Ollama (installed automatically during onboard)

---

📝 License

MIT © 2026 CharlesArea

---

Made with ❤️ for the AI community

GitHub • PyPI • Report Bug