Skip to content
entropy collection in JavaScript for your node and/or browser apps
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.
src version boost; even more (likely unnecessary) conservatism in entropy… Aug 28, 2014
test auto_stop of bit collection loop after each time generate is called Aug 16, 2014


The easiest way to generate good pseudorandom numbers in the browser is with window.crypto.getRandomValues, and in Node.js you can use crypto.rng. But for the truly paranoid, getting even more entropy is a good idea. For example, one might seed their own key generator with a combination of window.crypto and a series of coordinates collected from mouse movements or key mashes.

Even though the mouse movements of the user are not very random, it's extra noise, adding a layer of safety. Perhaps each [x,y] mouse location is worth a bit or two of entropy.

more-entropy achieves similar results but without user interaction or ugly integration with your DOM. It generates large numbers by counting how many operations it can perform in a unit of time, which fluctuates unpredictably based on other system processes and low-level architectural specifics (like cache misses and FPU pipelines).

A good use of this module is to combine its output with window.crypto.getRandomValue or crypto.rng, and use the result as a seed for a deterministic random bit generator (like HMAC_DRBG). You'll have an extra layer of protection if you're afraid that the standard random number generators are compromised.


npm install more-entropy


var m = require('more-entropy');

// create a generator, which can provide you with some entropy
var c = new m.Generator();

// get an array of integers with at least 100 bits of combined entropy:
c.generate(100, function(vals) {
  console.log(vals); // [-4358,543,9089,...]

What it's doing

This generator repeatedly does as many floating point operations as it can in 1ms-2ms time periods (typically many thousands), and compares this value to previous attempts. The delta is then added to a collection with a very conservative estimate for bits of entropy.

Much like the mouse movement technique, we are collecting a lot of data and assuming it's just a little bit random.


  • entropy is calculated by changes in performance; for example, extreme high performance with no variation yields zero entropy. Only fluctuations are captured.
  • this will work even if your system is bogged down (it'll just take longer)
  • it only CPU blocks for bursts up to 2ms, so it's safe in the browser and in Node.js
  • generate can be called as many times as you like, even concurrently; it will call back with uniquely calculated data to each request
  • return values are small integers (sometimes < 1000) and may be negative
  • entropy is collected over time, so a request for lots of bits could take a while

One Big Assumption

  • your CPU is not shared with an attacker; a carefully timed attack on the CPU could produce entropy less than what's requested


new m.Generator() can be called with extra options:

var c = new m.Generator({
  'loop_delay':        10 // how many milliseconds to pause between each operation loop. A lower value will generate entropy faster, but will also be harder on the CPU
  'work_min':           1 // milliseconds per loop; a higher value blocks the CPU more, so 1 is recommended
  'auto_stop_bits':  4096 // the generator prepares entropy for you before you request it; if it reaches this much unclaimed entropy it will stop working
  'max_bits_per_delta': 4 // a safety cap on how much entropy it can claim per value; 4 (default) is very conservative. a larger value will allow faster entropy generation
You can’t perform that action at this time.