Token-Oriented Object Notation (TOON): A token-efficient, schema-aware JSON for LLM prompts
Summary
TOON is a compact, schema-aware JSON encoding intended for LLM prompts, combining YAML-like indentation with CSV-style tabular arrays to reduce token usage while preserving structure. The README outlines benchmarks, features, installation, and tooling (CLI and TypeScript SDK), plus guidance on when TOON is advantageous or not for data delivery to models. It also highlights editor support, playgrounds, and adoption guidance for developers integrating TOON into AI workflows.