Toon Format

If you're hitting token limits or paying per-token on LLM APIs, this skill converts your JSON to TOON format, which cuts tokens by around 40% using YAML-style indentation for objects and CSV-style tables for arrays. It's genuinely more compact for uniform data like employee records or product catalogs. The encode/decode functions round-trip cleanly, and there's a CLI for quick conversions. Works well when you're injecting structured context into prompts or doing RAG retrieval where you have lots of similar objects. The format is readable enough that models understand it with a short system prompt explaining the bracket notation.