LilScript makes JavaScript libraries smaller
https://yeargun.github.io/lilscript/ LilScript is a typed, compression-first language that compiles into js and sometimes into exec(will be more stable in future). The compiler mangles, reshapes the program into optimized js that happens to be 5-15% smaller (after gzip/br compression or raw) compared to the best performing JS toolchains like oxc/esbuild/terser/.. ## What has been proven to work with LilScript? - makes VSCode's core js modules 20% smaller on average - makes the world's most performant & small markdown rendering npm library, marked, 5-7%…
In plain words
LilScript is a typed, compression-first programming language that compiles to JavaScript. It produces optimized code that is 5-15% smaller than standard JS toolchains like esbuild and terser, even after gzip compression. The tool is designed for developers building JavaScript libraries who want to reduce file size without sacrificing performance. It has demonstrated results on real projects, including reducing VSCode's core modules by 20% on average and improving both size and speed of existing npm packages.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
https://yeargun.github.io/lilscript/ LilScript is a typed, compression-first language that compiles into js and sometimes into exec(will be more stable in future). The compiler mangles, reshapes the program into optimized js that happens to be 5-15% smaller (after gzip/br compression or raw) compared to the best performing JS toolchains like oxc/esbuild/terser/.. ## What has been proven to work with LilScript? - makes VSCode's core js modules 20% smaller on average - makes the world's most performant & small markdown rendering npm library, marked, 5-7% smaller and 10% faster - and many more demos.. works with pretty much any js/ts library ## How it is compressing js finer than vite/oxc/terser/esbuild/.. ### 1. By changing the app. ``` class Vector { float x; float y; init(float x, float y) { this.x = x; this.y = y; } float lengthSquared() { return this.x * this.x + this.y * this.y; } } int[] values = [1, 2, 3, 4]; auto doubled = values.map((int value) => value * 2); int sum = 0; for (int i = 0; i < doubled.length; i++) { sum += doubled[i]; } Vector vector = new Vector(3.0, 4.0); if (vector.lengthSquared() == 25.0) { print(`sum=${sum}`); } ``` Above code compiles into this: ``` var b=[1,2,3,4].map(a=>a2|0); var a=0,c=0; while(a<b.length){ c=c+b[a]|0; a=a+1|0; } console.log(`sum=${c}`) ``` This is not minification of the same program. The compiler changed the app. oxc / esbuild / terser / .. starts from JS and mostly keeps the shape of the app. Meanwhile LilScript the language is designed from scratch to give compiler any extra knowledge that could help the compiler's compilation. Just like Google Closure Compiler - Advanced Mode, and beyond. ### 2. Tryhard property mangling. Visit the world's most used web applications, chatgpt.com, and read the source codes. You will see that, there are lots of framework related js properties that dont get minified, and are indeed human readable. Those names stay in the bundle because the toolchain cannot prove they are local. A property might be a public API, a DOM field, a framework hook, or something a plugin reads by string. So the minifier leaves it. LilScript: - *a-* does both eliminates the objects, weird code structures into more optimized variables/arrays - *b-* rename any variable into mostly occured, short versions, field cleverly to minimize entropy (based on the objective compression algorith. gzip/brotli) so that the end result is highly compressed. *(a)* is the same move as the `Vector` example, at library scale: objects and classes that only exist as a programming convenience get flattened into scalars, arrays, and tight loops. *(b)* is not "make every name 1 letter." gzip and brotli win when the same short tokens repeat. The compiler picks the names that show up the most, and scores the spelling against the compression algorithm you asked for (gzip, brotli, or raw). Different `cost_model` → different names → a different file that is smaller after* that codec. That is why the same program can be 5-15% smaller after gzip/br compression or raw: the JS is shaped and named for the compressor, not just for a human reading the AST. Feel free to PR, experiment (Please respect the modified MIT license) LLM models can oneshot implement your library with LilScript. For non optimized libraries, it could end up 20%+ size reduction and somewhat performance improvements (which usualy matters very little) Please share your opinions, would love to discuss about the future direction for the language and the compiler
More dev tools this month
the category →
Open-source GTM skills for technical founders
Dev tools · 29d ago · gtmcofounder.com
OpenTrailPaper is open-source bike computer firmware for the LilyGO T5S3 4.7" E-Paper PRO. It supports offline maps, GPX routes, FIT recording and Bluetooth sensors.
Dev tools · 1d ago · opentrailpaper.com
Launched alongside, August 2026
the whole month →- TL
Life & fun · 10d ago · louisabraham.github.io
- SA
Hello HN! I found that picking out plausible but diverse skin tones for my digital art and game development projects was kind of difficult, and I got curious about if there was a way to define a color space that made it easy. I've built a color picker and procedural generation algorithm based on the space as well as a bunch of other fun js features and demos throughout the page that use the equations. If you find it interesting, I have lots of explanations of how I built it and what properties the space has. The methodology might be a bit shaky, but hopefully the result is as helpful for…
Life & fun · Aug 2026 · toneyalexander.github.io
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 16d ago · simedw.com