Surprise! 🎉 I’m releasing a new data utility library today.
Meet ** @srtv/toondash** — a clean, declarative, and framework-agnostic set of utilities built to query, slice, and transform data payloads directly on the wire without full object inflation overhead.
When you pass large structural datasets (like user lists, analytics logs, or product catalogues) into an LLM context window, standard JSON introduces an incredible amount of syntactic noise. You waste up to 40% of your prompt token budget repeating dictionary keys, quotes, and braces.
Newer text formats like Token-Oriented Object Notation (TOON) fix this by stripping punctuation and declaring schemas once, slashing your token footprints by 40% to 70%.
However, there are enough subtle "gotchas" with compressed data streams that add up to an annoying amount of complexity in production:
We need data compression, but managing the data shouldn't be a mess.
@srtv/toondash
@srtv/toondash
provides a unified processing layer. It doesn't care about your format. It natively ingests BOTH standard JSON data and compressed TOON string layouts through a single rich query API, handling the underlying token compression and structural recalculations automatically under the hood.
import { filter, map } from '@srtv/toondash';
// Input standard JSON arrays OR compressed TOON text
// It intelligently runs the query and outputs the compressed stream seamlessly
const optimizedStream = filter(anyInputFormat, { status: 'Active' });
filter
, map
, pick
) that integrate cleanly into any Node.js or JavaScript AI workflow.Skip the local setup entirely. I have launched a live interactive sandbox where you can drop your raw data payloads in and watch real-time multi-format mutations and compression take place:
👉 Launch the Live ToonDash Playground
Give it a spin via your terminal today:
npm install @srtv/toondash
This is the initial release, and I am aiming to continuously scale out the supported utility methods as production workflows become more complex. Check out the project, grab the code, and let me know if you run into any bugs or have feature feedback on our ** Official Documentation Hub**!