The universal token compressor that gives your wallet (and your AI Agents) room to breathe

If you develop with LLMs, you know the drama: agent loops that continuously repeat the same context, gigantic JSON files that clog the context window, and those end-of-month API bills that send shivers down your spine.

That's why I decided to launch Synthelion — a universal Token Compressor and Prompt Manager designed to cut token usage by up to 70%, speeding up model responses without losing an ounce of meaning.

And yes, there's a quote I couldn't resist:

"Why use many tokens when few tokens do same job?" — A Neanderthal man (and your wallet).

🔥 What is Synthelion and how does it work?

Synthelion interposes itself between your code (or your agent) and the LLM. It analyzes the prompt, removes all superfluous "grammatical packaging" (articles, prepositions, conjunctions), and reduces words to their base form (lemmatization). The AI receives the exact same essential information, but distilled.

The Before and After (Some practical examples)

1. Prose in Italian (Savings ~52%):
2. Before: I would like to know if it's possible to receive information about cheap restaurants in Rome, please.
3. After: know possible receive information cheap restaurant Rome
4. JSON Array (Savings ~69%):
5. Before: A classic JSON with repeated keys for each object.
6. After: It is automatically converted into a clean and lossless Markdown table, which models natively digest with a fraction of the tokens.
7. HTML Pages, Build Logs, and Git Diffs:
8. Synthelion integrates an intelligent Content Router: it understands what you are passing to it (code, logs, tables, or text) and applies the perfect compression algorithm for that format.
9. 
   

🛠️ Universal Integration (Zero configuration)

Synthelion was born to be integrated anywhere in 5 minutes:

1. 🔌 MCP Protocol: Native support for Claude Code, Cursor, Windsurf, and Claude Desktop. You just need to add a line to the configuration file to give your IDE compression superpowers.
2. 🐍 OpenAI & LangChain Plugin: Ready-to-use tools to be passed as tools to the OpenAI API or in your ReAct agents on LangGraph.
3. 💻 CLI Interface and Python API: You can use it directly from the terminal in Bash pipelines or integrate it into your scripts with pip install synthelion.

🌍 Sustainability and Numbers at Hand

It's not just a matter of costs (which, on volumes of 10M tokens per day, mean thousands of dollars saved per year), but also of environmental impact.

Synthelion includes an integrated energy estimator: every token saved avoids approximately 0.005 mWh of computing energy and 0.002 mg of CO₂. You can track efficiency directly from your code:

Python

📦 Where to start?

The project is open-source, supports more than 50 languages out-of-the-box (automatically detected), and does not require any AI model to run locally. It is the natural evolution of a previous C# project of mine (Caveman), redesigned from scratch for the Python and AI Agents ecosystem.

1. 🐍 PyPI: pip install synthelion
2. 🛠️ GitHub Repository: francescopaolopassaro/synthelion