🦍 GOON

Greatly Optimized Object Notation

"Edging the competition, one token at a time."

A hyper-compact, LLM-optimized data serialization format. GOON achieves 37.9% token savings vs JSON and 92.5% LLM accuracy – beating TOON on both metrics.

Installation

npm install @goon-format/goon

Quick Start

import { encode, decode } from '@goon-format/goon';

const data = {
  users: [
    { id: 1, name: 'Alice', role: 'admin', active: true },
    { id: 2, name: 'Bob', role: 'user', active: true },
    { id: 3, name: 'Carol', role: 'admin', active: false },
  ]
};

// Encode (defaults to LLM mode for best accuracy)
const goon = encode(data);
console.log(goon);
// users[3]{id,name,role,active}:
// 1,Alice,admin,T
// 2,Bob,user,T
// 3,Carol,admin,F

// Decode back to JavaScript
const restored = decode(goon);

Mode Presets

// 🧠 LLM Mode (default) - Best for sending to GPT/Claude
encode(data);  // or encode(data, { mode: 'llm' })

// 🍆 Compact Mode - Maximum compression
encode(data, { mode: 'compact' });

// ⚖️ Balanced Mode - Readable with compression
encode(data, { mode: 'balanced' });

Key Features

• Tabular Arrays: Objects with same keys become CSV-like tables
• Compact Literals: T/F/_/~ for true/false/null/empty
• Mode Presets: llm, compact, balanced
• Optional Dictionary: String deduplication with $0, $1 refs
• Optional Column Refs: ^ means "same as above"

Benchmarks

| Metric | JSON | TOON | GOON | |--------|------|------|------| | Token savings | - | 36.5% | 37.9% | | LLM Accuracy | 80.8% | 86.8% | 92.5% | | Efficiency Score | 11.3 | 18.0 | 22.3 |

API

encode(value, options?)

encode(data, {
  mode: 'llm',           // 'llm' | 'compact' | 'balanced'
  dictionary: false,     // String deduplication
  columnRefs: false,     // ^ references
  minimalIndent: true,   // Remove row indentation
});

decode(input, options?)

decode(goonString, {
  reviver: (key, value) => value,
});

Security

• Prototype pollution prevention
• Recursion depth limits (100 levels)
• Input size limits (10MB)
• RLE expansion limits (10,000)

Links

• GitHub
• Full Documentation
• SPEC

License

MIT