Syntax Cheatsheet

Basic Types

Primitives

# String (unquoted when safe)
name:Alice

# Number
score:98.5
count:42

# Boolean (T/F)
active:T
disabled:F

# Null
value:null

Objects

# Simple object
name:ZON Format
version:1.0.5
active:T
score:98.5

JSON equivalent:

{
  "name": "ZON Format",
  "version": "1.0.5",
  "active": true,
  "score": 98.5
}

Nested Objects

Colon-less Syntax (v2.0.5):

# Colon is optional if value starts with { or [
config{database{host:localhost,port:5432},cache{ttl:3600,enabled:T}}

Legacy Quoted (v1.x):

config:"{database:{host:localhost,port:5432},cache:{ttl:3600,enabled:T}}"

JSON equivalent:

{
  "config": {
    "database": { "host": "localhost", "port": 5432 },
    "cache": { "ttl": 3600, "enabled": true }
  }
}

Arrays

Primitive Arrays (Inline)

tags"[nodejs,typescript,llm]"
numbers"[1,2,3,4,5]"
flags"[T,F,T]"

Tabular Arrays (Uniform Objects)

Most efficient form - ZON's specialty

users:@(3):active,id,name,role
T,1,Alice,admin
T,2,Bob,user
F,3,Carol,guest

Breakdown:

• @(3) = 3 rows
• :active,id,name,role = column headers
• Data rows follow

JSON equivalent:

{
  "users": [
    { "id": 1, "name": "Alice", "role": "admin", "active": true },
    { "id": 2, "name": "Bob", "role": "user", "active": true },
    { "id": 3, "name": "Carol", "role": "guest", "active": false }
  ]
}

Mixed / Non-Uniform Arrays

# Quoted as JSON array
items"[{id:1,name:Alice},{id:2,name:Bob,role:admin}]"

Empty Containers

# Empty object
metadata"{}"

# Empty array
tags"[]"

Quoting Rules

When Strings NEED Quotes

1. Contains special characters:

   • Commas: "hello, world"
   • Colons: "key:value"
   • Brackets: "[test]"
   • Braces: "{test}"

2. Looks like a literal:

   • "true" (string, not boolean)
   • "123" (string, not number)
   • "false" (string, not boolean)
   • "null" (string, not null)

3. Leading/trailing spaces:

   • " padded "

4. Empty string:

   • "" (MUST quote, otherwise parses as null)

Safe Unquoted Strings

# Alphanumeric + dash, underscore, dot
name:john-doe
file:data_v1.json
host:api.example.com

Table Headers

Basic Header (without count)

users:@:id,name,active
1,Alice,T
2,Bob,F

Full Header (with count)

users:@(2):id,name,active
1,Alice,T
2,Bob,F

Best practice: Always include count @(N) for explicit schema

Type Conversions

Common Patterns

Config with nested data

environment:production
version:"1.0.5"
database:"{host:db.example.com,port:5432,ssl:T}"
features:"{darkMode:F,betaAccess:T}"

Mixed structure (tables + metadata)

created:"2025-11-28"
total:150
users:@(3):id,name,status
1,Alice,active
2,Bob,inactive
3,Carol,active

Root-level table

@(2):id,name,active
1,Alice,T
2,Bob,F

Escape Sequences

Within quoted strings:

• \" - Double quote
• \\ - Backslash
• \n - Newline
• \r - Carriage return
• \t - Tab

Example:

message:"Line 1\nLine 2"
path:"C:\\Users\\data"

Complete Example

JSON:

{
  "metadata": { "version": "1.0.5", "env": "production" },
  "users": [
    { "id": 1, "name": "Alice", "active": true, "loginCount": 42 },
    { "id": 2, "name": "Bob", "active": true, "loginCount": 17 },
    { "id": 3, "name": "Carol", "active": false, "loginCount": 3 }
  ],
  "config": { "database": { "host": "localhost", "port": 5432 } }
}

ZON:

metadata{version:1.0.5,env:production}
users:@(3):active,id,loginCount,name
T,1,42,Alice
T,2,17,Bob
F,3,3,Carol
config.database{host:localhost,port:5432}

Token count:

• JSON (formatted): 151 tokens
• ZON: 87 tokens
• Savings: 42% fewer tokens

Tips for LLMs

1. Use code blocks: Wrap ZON in ```zon for syntax highlighting
2. No hints needed: Format is self-documenting
3. Explicit headers: @(N) count helps LLMs validate
4. Column names: Listed once, clear schema

Example prompt:

Here's the data in ZON format:

```zon
users:@(3):id,name,active
1,Alice,T
2,Bob,F
3,Carol,T
```

Question: How many active users are there?

Quick Comparison

See also:

• Format Specification - Formal grammar
• API Reference - encode/decode functions
• LLM Best Practices - Usage guide