Neak Peek: Bringing TOON Format to My Upcoming JSON Multi-Tool
If you regularly feed structured data into Large Language Models (LLMs) like Claude or DeepSeek, you know that tokens equal money and context limits. Standard JSON is highly readable for machines, but all those repetitive keys, quotes, and brackets waste a massive amount of valuable token space.
To solve this inefficiency in my own AI workflows, I am currently building a client-side JSON Multi-Tool. Alongside standard formats like CSV and XML, I am particularly excited to integrate TOON (Token-Oriented Object Notation).
What is TOON?
TOON is an emerging data format optimized specifically for LLM prompts. It merges the clean indentation of YAML with the tabular efficiency of CSV. By stripping away redundant syntax, it can compress data payloads by 30% to 60%.
Here is a quick look at the transformation. A standard JSON payload:
{
"ecosystem": "aslando",
"services": [
{ "id": 1, "name": "pac-kube", "type": "game" },
{ "id": 2, "name": "focus", "type": "app" }
]
}
Will be instantly flattened by the upcoming tool into highly compressed TOON notation:
ecosystem: aslando
services[2]{id,name,type}:
1,pac-kube,game
2,focus,app
I have been wrapping up the core JavaScript conversion engine for this utility. It will allow you to quickly optimize logs, configs, and datasets for your AI prompts—running 100% locally in your browser for absolute data privacy.
Happy coding, everyone!