JTON Playground

Fast JSON-compatible parsing with JTON encoding, token insights, and speed context.

Paste JSON, tune JTON output options, compare token footprint instantly, and copy an LLM-ready format hint. It now works in two modes: the real Python backend when available, and a browser-only fallback so the same page can be published on GitHub Pages.

Encoding...

Input size 0

characters

Output size 0

characters

JTON encode time -

auto runtime

Savings -

vs compact JSON

Round-trip Waiting

encode and parse-back check

JSON Input

Paste JSON or load one of the sample datasets.

JTON Output

Live serialized output using your selected options.

Token lens

Color highlights show the rough token makeup of the current JTON output so users can quickly spot bulky strings, repeated identifiers, and punctuation-heavy segments.

Token comparison

Current output versus JSON baselines, plus TOON as a comparison format that is not JSON-compatible.

Output controls

Tune JTON for readability, compactness, or LLM usage.

Structure

Zen Grid Convert tabular arrays of objects into compact JTON table form.

Include row count Add the structural [N: prefix. Recommended for LLM retrieval.

Encoding

Bare strings Use unquoted identifier-like values where safe to reduce clutter.

Implicit null Represent null cells as empty values for maximum compactness.

Unquoted keys Emit object keys in JTON style without JSON quotes.

Delimiter

Formatting

LLM prompt helper

Verified against core: these prompt primer styles still exist in jton.format_hint(...).

Hosting

This page can run as a static GitHub Pages site. If the API is unavailable, the browser fallback keeps encode, decode, hints, and token estimates working.

Core sync

Playground controls are aligned with the real API: row_count, bare_strings, implicit_null, delimiter=comma|tab|pipe, and prompt hint styles.

Dataset token and stdlib-json speed snapshot

Live speed cards below compare Python stdlib json parse/dump against JTON on the current JSON input. The dataset table remains a broader benchmark-style snapshot across flat, nested, and semi-uniform data. TOON is shown for comparison only and is not JSON-compatible.

Stdlib json parse - Current input, backend benchmark

JTON parse - Current input, backend benchmark

Stdlib json dump - Current input, backend benchmark

JTON dump - Current input, backend benchmark

Dataset	Shape	JSON compact	TOON (not JSON)	JTON row-count	JTON tab	Best savings	Visual

Sample datasets

Try realistic structures to see how JTON behaves across flat, nested, and semi-structured data.

Spec snapshot

JTON is a JSON-superset encoding with a tabular mode for repeated objects. It stays readable while reducing tokens for flat datasets.

[3: id, name, role; 1, "Ada", "admin"; 2, "Bob", "user"; 3, "Cara", "ops" ]

LLM-friendly defaults

Row count is enabled by default because it helps structural awareness. Use bare strings and implicit nulls only when you want maximum compactness.

row_count=true, zen_grid=true, delimiter="comma"

Static deployment

This page can be hosted on GitHub Pages. When the Python API is unavailable, it automatically switches to the browser runtime for encode, decode, hints, and token estimates.

local backend -> Python mode | github.io -> static fallback