Random Number Generator (cryptographically secure, unique mode)

Why this is actually random

We use crypto.getRandomValues — the browser's cryptographically secure random source — instead of Math.random(). The CSPRNG output can't be predicted from prior values, which matters for lottery picks, gambling-adjacent applications, and any use where 'random' is a real requirement, not just 'arbitrary'.

We also use rejection sampling to make the distribution exactly uniform across the chosen range. Naive 'random % range' introduces bias when the range doesn't divide evenly into 2³². Rejection sampling discards values that would cause bias and tries again, so every number in your range has exactly the same probability.

Unique vs duplicate-allowed

Default mode: each number is independent — duplicates can occur. This is what 'random' usually means in casual use (rolling dice, picking a movie, randomized testing).

Unique mode: sample without replacement — once a number is picked, it can't be picked again in the same batch. Useful for: lottery picks (e.g., 6 numbers from 1-49), raffles (drawing winners), team assignments, and any 'pick N from M' situation. Requires count ≤ range.

Common use cases

Lottery / raffle: 6 unique numbers from 1-49, or whatever your draw uses. Toggle 'unique' on.

Dice rolls: range 1-6 (or 1-20 for D20). Leave 'unique' off so each roll is independent.

Random sampling for testing: pick 10 random user IDs from 1-10000 to spot-check. Use 'unique' to avoid duplicates.

Simulating coin flips: range 0-1, multiple results. Tally heads/tails for fairness checks (you'll see ~50/50 on large samples).