BIP-39 Splitter

bip39-splitter is a command line tool to split (and reconstruct) a BIP-39 mnemonic (seed phrase). If done correctly, this improves the availability and security of your mnemonic by adding fault tolerance and decentralization in the storage of your mnemonic.

bip39-splitter is:

• Cross-platform: it will compile to all OSes/architectures supported by Golang
• Self-contained: a single binary with no dependency to make sure it will always work
• Small: <400 lines of code excluding the Shamir Secret Sharing implementation (~600 lines including it)

All languages officially supported by BIP-39 (as of March 27, 2025) are supported:

• English
• Japanese
• Korean
• Spanish
• Chinese (Simplified)
• Chinese (Traditional)
• French
• Italian
• Czech
• Portuguese

Dictionaries were copied from the official BIPs repository.

Usage

Important

For best security, this program should be compiled and executed on an air-gapped computer.

First, build the executable:

go build -o bip39splitter cmd/bip39splitter/main.go

Then, split your mnemonic:

$ ./bip39splitter split -l english -p 5 -t 3
Enter your BIP-39 mnemonic (seed phrase): icon thunder tube include demise charge valley odor asthma volcano cost invest again hidden fiction goose bulk twin retire chair tooth genius flight erase

Parts:
==================================================================================================
caff12146c4ba97dd69ed64752d44a16903a01d8591d5d1cd47a653c89a47f8d3044259c0f993122786178f58f7d434cd9
9c5102654e850ea4c847ae14fc0ef3f85de3553d277affb369b17ca7152398b858b0fc276d3669908b4f0e07cd98be581b
a036298481395175e340563af8ec29a15327c71a0a05967dbb927a0b629fc1369b75a3686302e44487ffaeb632dd7eec6b
6bef4382cf502d0cc672227243ea71ef83661438736b092c1b118d9b6e1e6ac20abade05ecb22cf0e4a5dab7704fe3ede5
c6f475900b1c147918d5f1a4660a74d6f9bfa9aa88aeaafdb674354741a7cf89411d0e596c0723394d1a96dc8d47da8bcb
==================================================================================================

In this example, language is set to English (-l english), number of parts is set to 5 (-p 5) and number of required parts to be able to reconstruct the mnemonic is set to 3 (-t 3).

Tip

It is expected that the same mnemonic will give different parts each time as Shamir Secret Sharing is based on randomly generated numbers.

Later, you can reconstruct the mnemonic by gathering any 3 of the 5 parts (order doesn't matter):

$ ./bip39splitter combine -l english
Enter part 1 and press ENTER (leave empty to finish): 9c5102654e850ea4c847ae14fc0ef3f85de3553d277affb369b17ca7152398b858b0fc276d3669908b4f0e07cd98be581b
Enter part 2 and press ENTER (leave empty to finish): 6bef4382cf502d0cc672227243ea71ef83661438736b092c1b118d9b6e1e6ac20abade05ecb22cf0e4a5dab7704fe3ede5
Enter part 3 and press ENTER (leave empty to finish): caff12146c4ba97dd69ed64752d44a16903a01d8591d5d1cd47a653c89a47f8d3044259c0f993122786178f58f7d434cd9
Enter part 4 and press ENTER (leave empty to finish):

Reconstructed mnemonic (seed phrase):
icon thunder tube include demise charge valley odor asthma volcano cost invest again hidden fiction goose bulk twin retire chair tooth genius flight erase

Motivation

Splitting your mnemonic into multiple parts (stored in different places) brings two major availability and security advantages over storing it in one single place:

• Fault tolerance: depending on the splitting parameters you choose, you can afford to lose one or more parts of your mnemonic without compromising your ability to recover the mnemonic.
• Decentralization: an attacker needs to gather several parts of your mnemonic (requiring access to different places) to steal it, making it much more difficult than if it is stored wholly in a single place.

In other words, splitting your mnemonic makes it more secure against attacks and your own mistakes.

Algorithm

There are some steps involved before and after splitting the secret with Shamir Secret Sharing. Here is the complete algorithm implemented in bip39-splitter:

1. Turn each word of the mnemonic into its corresponding index in the BIP-39 dictionary of the chosen language (e.g. word ability of the English dictionary becomes 2, as it's the second word of the list).
2. Convert word indices to an array of bytes (keep the order!). Indices are 2 bytes long and encoded in Big Endian (e.g. index 2 from the word ability is represented as 0x00 0x02).
3. Apply Shamir Secret Sharing to the resulting array of bytes. The SSS implementation used is the one present in Hashicorp Vault v1.19.0.
4. Encode each part in its hexadecimal string representation (see Golang's hex.EncodeToString() function implementation for details).

Shamir Secret Sharing (SSS) implementation

As per the concept of Don't roll your own cryptography, the SSS implementation (which is the core of this tool) was copied as-is from Hashicorp Vault, an enterprise-grade secrets management software. The version used in bip39-splitter is extracted from Vault v1.19.0.

The code was copied (instead of imported as a Go module) in order to avoid any build dependency, as it is critical to be able to (re)build this software many years from now.

Important

As all implementations of Shamir Secret Sharing tend to be slightly different (plus the fact that we add extra steps besides SSS), this software is not compatible with other SSS implementations such as the sss tool on Linux. This means YOU NEED TO KEEP A COPY OF THIS PROGRAM TO RECONSTRUCT YOUR MNEMONIC IN THE FUTURE.

Licensing

This software is distributed under two licenses:

• The Shamir Secret Sharing implementation, being taken from Hashicorp Vault, remains under their Mozilla Public License version 2.0.
• The rest of the program is under MIT License, as permitted by the Mozilla Public License version 2.0.