TOON: Token-Oriented Object Notation - A Compact JSON Format for LLM Prompts
By
royosherove
Pulled from the oven just right. Trustworthy, fact-dense, deeply satisfying.
Summary
TOON (Token-Oriented Object Notation) is a new data format designed specifically for LLM prompts that provides a compact, human-readable encoding of JSON data. It combines YAML's indentation-based structure for nested objects with CSV-style tabular layout for uniform arrays, minimizing token usage while maintaining lossless JSON representation. The format is particularly optimized for uniform arrays of objects, achieving CSV-like compactness while adding explicit structure to help LLMs parse and validate data reliably. The article presents the specification, benchmarks, and a TypeScript SDK for this new data format.
Key quotes
· 4 pulledToken-Oriented Object Notation is a compact, human-readable encoding of the JSON data model that minimizes tokens and makes structure easy for models to follow.
TOON combines YAML's indentation-based structure for nested objects with a CSV-style tabular layout for uniform arrays.
TOON's sweet spot is uniform arrays of objects (multiple fields per row, same structure across items), achieving CSV-like compactness while adding explicit structure that helps LLMs parse and validate data reliably.
It's intended for LLM input as a drop-in, lossless representation of your existing JSON.
You might also wanna read
whichllm: An open-source CLI tool that ranks local LLMs by real benchmarks and hardware compatibility
whichllm is an open-source CLI tool that auto-detects your GPU/CPU/RAM specs and ranks the best local LLMs from HuggingFace that actually ru
github.com·16d agoGeoJSON: A Standardized Format for Geographic Data Encoding (RFC 7946)
GeoJSON is a standardized format (RFC 7946) for encoding geographic data structures. It supports geometry types including Point, LineString,
Intel's AutoRound: Open-Source Quantization Toolkit for Low-Bit LLM and VLM Inference
AutoRound is an advanced quantization toolkit for Large Language Models (LLMs) and Vision-Language Models (VLMs) developed by Intel. It achi
github.com·1mo ago
cc-canary: Offline Drift Detection Tool for Claude Code
A pre-alpha drift detection tool for Claude Code, packaged as two Agent Skills. It reads existing JSONL session logs from Claude Code to det
github.com·1mo ago
Respan: Self-Driving AI Observability and Evaluation Platform for LLM Agents
Respan is a DevOps platform for LLM applications that provides self-driving AI observability and evaluation tools for AI agents. With just 2
Product Hunt·1mo ago
Seeknal: A CLI Tool for Data & AI/ML Pipelines with Natural Language Queries
Seeknal is a CLI tool for data and AI/ML engineering that allows users to define pipelines in YAML or Python, run a safe draft→dry-run→apply
Product Hunt·1mo ago