TOON Format for Go

Token-Oriented Object Notation is a compact, human-readable format designed for passing structured data to Large Language Models with significantly reduced token usage.

Example

JSON (verbose):

{
  "users": [
    { "id": 1, "name": "Alice", "role": "admin" },
    { "id": 2, "name": "Bob", "role": "user" }
  ]
}

TOON (compact):

users[2]{id,name,role}:
  1,Alice,admin
  2,Bob,user

Usage

Marshal and Unmarshal

package main

import (
    "fmt"

    "github.com/toon-format/toon-go"
)

type User struct {
    ID    int    `toon:"id"`
    Name  string `toon:"name"`
    Role  string `toon:"role"`
}

type Payload struct {
    Users []User `toon:"users"`
}

func main() {
    in := Payload{
        Users: []User{
            {ID: 1, Name: "Alice", Role: "admin"},
            {ID: 2, Name: "Bob", Role: "user"},
        },
    }

    encoded, err := toon.Marshal(in, toon.WithLengthMarkers(true))
    if err != nil {
        panic(err)
    }
    fmt.Println(string(encoded))

    var out Payload
    if err := toon.Unmarshal(encoded, &out); err != nil {
        panic(err)
    }
    fmt.Printf("first user: %+v\n", out.Users[0])
}

Unmarshal into Maps

Unmarshal can populate dynamic maps, mimicking the encoding/json package:

var doc map[string]any
if err := toon.Unmarshal(encoded, &doc); err != nil {
    panic(err)
}
fmt.Printf("users: %#v\n", doc["users"])

Decode Without Structs

If you do not have a destination struct, use Decode for a dynamic representation:

package main

import (
    "fmt"
    "github.com/toon-format/toon-go"
)

func main() {
    raw := []byte("users[2]{id,name,role}:\n  1,Alice,admin\n  2,Bob,user\n")
    decoded, err := toon.Decode(raw)
    if err != nil {
        panic(err)
    }
    fmt.Printf("%+v\n", decoded)
}

For more runnable samples, explore the programs in ./examples.

Resources

• TOON Specification
• Main Repository
• Benchmarks & Performance
• Other Language Implementations

Contributing

Interested in implementing TOON for Go? Check out the specification and feel free to contribute!

License

MIT License © 2025-PRESENT Johann Schopplich