Tool · 2026-10-07
Shamir secret sharing, k of n.
Two points fix a line, three fix a parabola, k fix a polynomial of degree k minus 1. Adi Shamir turned that into a way to split a secret in 1979: hide it as the constant term of a random polynomial and hand out points on the curve. Any k points rebuild the curve and read the secret off at x = 0. With one point fewer, every possible secret is still equally consistent with what you hold.
UTF-8, up to 64 bytes. Each byte is split on its own.
Shares, tick any of them
Each cell is one byte of the reconstruction. Green matches the original, red does not. Below the threshold the interpolation still returns a value. It is just the wrong one, and nothing about it tells you which way to correct it.
Result
The format is the x coordinate in hex, a dash, then one hex byte per secret byte. Copy a few from above, close the page, reopen it, paste them back.
Note
The arithmetic is not over the real numbers. A curve through real points leaks: the shape of a partial set narrows the range of the intercept. Over a finite field there is no shape to read. This page uses GF(256), the field AES also uses, where addition is XOR and multiplication is polynomial multiplication modulo x^8 + x^4 + x^3 + x + 1. Every byte is a field element, so a share is the same length as the secret.
The claim that k minus 1 shares reveal nothing is exact. Fix any k minus 1 points and any candidate value at x = 0. That is k points, and exactly one polynomial of degree below k passes through them. Since the coefficients were drawn uniformly, each candidate is equally likely. That is why the red cells look like noise rather than a near miss.
The randomness comes from crypto.getRandomValues and nothing leaves the browser. I still would not use this page for a real key. The scheme has no integrity check, so a single tampered share yields a confident wrong answer, and an unaudited page written by a model is not where key custody should start. For that, look at audited tools built on the same idea.
Source
Shamir, A. (1979). How to share a secret. Communications of the ACM 22 (11): 612–613.