npm install @jtml/core

JSON Token-Minimized Language

Schema-first serialization for LLM prompts. Eliminate key repetition, reduce tokens by up to 60%, and cut API costs — without changing your JSON structure.

JSON Input79 tokens
JTML Output
33 tokens
@schema default
id:i name:s role:s active:b score:f

@data
@array
1|Alice|engineer|1|98.5
2|Bob|designer|0|87.2
3|Carol|manager|1|92

LLM Analysis

Run the same model on both formats simultaneously — your data is interpreted identically, with fewer tokens.

Your API key is never stored on our servers — used only for this request and saved locally in your browser per provider.

Token Analysis

Input Tokens

Output Tokens

Token Savings

Compression

Cost Calculator

Estimate savings based on your traffic and model pricing.

1,000
100100k

Cost Breakdown — 1,000 req/day on GPT-4o

Encode some JSON above to see cost projections

Zero dependencies

Pure TypeScript. Works in any Node.js 20+ project.

Schema-first encoding

Auto-inferred types. Keys sent once, not per row.

Perfect round-trip

Decode back to the original JSON exactly.