noble-hashes

Audited & minimal JS implementation of hash functions, MACs and KDFs.

• 🔒 Audited by an independent security firm
• 🔻 Tree-shakeable: unused code is excluded from your builds
• 🏎 Fast: hand-optimized for caveats of JS engines
• 🔍 Reliable: chained / sliding window / DoS tests and fuzzing ensure correctness
• 🔁 No unrolled loops: makes it easier to verify and reduces source code size up to 5x
• 🦘 Includes SHA, RIPEMD, BLAKE, HMAC, HKDF, PBKDF, Scrypt, Argon2 & KangarooTwelve
• 🪶 48KB for everything, 4.8KB (2.36KB gzipped) for single-hash build

Take a glance at GitHub Discussions for questions and support. The library's initial development was funded by Ethereum Foundation.

This library belongs to noble cryptography

noble cryptography — high-security, easily auditable set of contained cryptographic libraries and tools.

• Zero or minimal dependencies
• Highly readable TypeScript / JS code
• PGP-signed releases and transparent NPM builds
• All libraries: ciphers, curves, hashes, post-quantum, 4kb secp256k1 / ed25519
• Check out homepage for reading resources, documentation and apps built with noble

Usage

npm install @noble/hashes

deno add jsr:@noble/hashes

deno doc jsr:@noble/hashes # command-line documentation

We support all major platforms and runtimes. For React Native, you may need a polyfill for getRandomValues. A standalone file noble-hashes.js is also available.

1// import * from '@noble/hashes'; // Error: use sub-imports, to ensure small app size
2import { sha256 } from '@noble/hashes/sha2.js'; // ESM & Common.js
3sha256(Uint8Array.from([0xca, 0xfe, 0x01, 0x23])); // returns Uint8Array
4
5// Available modules
6import { sha256, sha384, sha512, sha224, sha512_224, sha512_256 } from '@noble/hashes/sha2.js';
7import { sha3_256, sha3_512, keccak_256, keccak_512, shake128, shake256 } from '@noble/hashes/sha3.js';
8import { cshake256, turboshake256, kmac256, tuplehash256, k12, m14, keccakprg } from '@noble/hashes/sha3-addons.js';
9import { blake3 } from '@noble/hashes/blake3.js';
10import { blake2b, blake2s } from '@noble/hashes/blake2.js';
11import { blake256, blake512 } from '@noble/hashes/blake1.js';
12import { sha1, md5, ripemd160 } from '@noble/hashes/legacy.js';
13import { hmac } from '@noble/hashes/hmac.js';
14import { hkdf } from '@noble/hashes/hkdf.js';
15import { pbkdf2, pbkdf2Async } from '@noble/hashes/pbkdf2.js';
16import { scrypt, scryptAsync } from '@noble/hashes/scrypt.js';
17import { argon2d, argon2i, argon2id } from '@noble/hashes/argon2.js';
18import * as utils from '@noble/hashes/utils'; // bytesToHex, bytesToUtf8, concatBytes...

• sha2: sha256, sha384, sha512
• sha3: FIPS, SHAKE, Keccak
• sha3-addons: cSHAKE, KMAC, K12, M14, TurboSHAKE
• blake, blake2, blake3 | legacy: sha1, md5, ripemd160
• MACs: hmac | sha3-addons kmac | blake3 key mode
• KDFs: hkdf | pbkdf2 | scrypt | argon2
• utils
• Security | Speed | Contributing & testing | License

Implementations

Hash functions:

• sha256(): receive & return Uint8Array
• sha256.create().update(a).update(b).digest(): support partial updates
• blake3.create({ context: 'e', dkLen: 32 }): sometimes have options
• support little-endian architecture; also experimentally big-endian
• can hash up to 4GB per chunk, with any amount of chunks

sha2: sha256, sha384, sha512 and others

1import { sha224, sha256, sha384, sha512, sha512_224, sha512_256 } from '@noble/hashes/sha2.js';
2const res = sha256(Uint8Array.from([0xbc])); // basic
3for (let hash of [sha256, sha384, sha512, sha224, sha512_224, sha512_256]) {
4  const arr = Uint8Array.from([0x10, 0x20, 0x30]);
5  const a = hash(arr);
6  const b = hash.create().update(arr).digest();
7}

See RFC 4634 and the paper on truncated SHA512/256.

sha3: FIPS, SHAKE, Keccak

1import {
2  keccak_224, keccak_256, keccak_384, keccak_512,
3  sha3_224, sha3_256, sha3_384, sha3_512,
4  shake128, shake256,
5} from '@noble/hashes/sha3.js';
6for (let hash of [
7  sha3_224, sha3_256, sha3_384, sha3_512,
8  keccak_224, keccak_256, keccak_384, keccak_512,
9]) {
10  const arr = Uint8Array.from([0x10, 0x20, 0x30]);
11  const a = hash(arr);
12  const b = hash.create().update(arr).digest();
13}
14const shka = shake128(Uint8Array.from([0x10]), { dkLen: 512 });
15const shkb = shake256(Uint8Array.from([0x30]), { dkLen: 512 });

See FIPS-202, Website.

Check out the differences between SHA-3 and Keccak

sha3-addons: cSHAKE, KMAC, K12, M14, TurboSHAKE

1import {
2  cshake128, cshake256,
3  k12,
4  keccakprg,
5  kmac128, kmac256,
6  m14,
7  parallelhash256,
8  tuplehash256,
9  turboshake128, turboshake256
10} from '@noble/hashes/sha3-addons.js';
11const data = Uint8Array.from([0x10, 0x20, 0x30]);
12const ec1 = cshake128(data, { personalization: 'def' });
13const ec2 = cshake256(data, { personalization: 'def' });
14const et1 = turboshake128(data);
15const et2 = turboshake256(data, { D: 0x05 });
16 // tuplehash(['ab', 'c']) !== tuplehash(['a', 'bc']) !== tuplehash([data])
17const et3 = tuplehash256([utf8ToBytes('ab'), utf8ToBytes('c')]);
18// Not parallel in JS (similar to blake3 / k12), added for compat
19const ep1 = parallelhash256(data, { blockLen: 8 });
20const kk = Uint8Array.from([0xca]);
21const ek10 = kmac128(kk, data);
22const ek11 = kmac256(kk, data);
23const ek12 = k12(data);
24const ek13 = m14(data);
25// pseudo-random generator, first argument is capacity. XKCP recommends 254 bits capacity for 128-bit security strength.
26// * with a capacity of 254 bits.
27const p = keccakprg(254);
28p.feed('test');
29const rand1b = p.fetch(1);

• Full NIST SP 800-185: cSHAKE, KMAC, TupleHash, ParallelHash + XOF variants
• Reduced-round Keccak:
  • 🦘 K12 aka KangarooTwelve
  • M14 aka MarsupilamiFourteen
  • TurboSHAKE
• KeccakPRG: Pseudo-random generator based on Keccak

blake, blake2, blake3

1import { blake224, blake256, blake384, blake512 } from '@noble/hashes/blake1.js';
2import { blake2b, blake2s } from '@noble/hashes/blake2.js';
3import { blake3 } from '@noble/hashes/blake3.js';
4
5for (let hash of [
6  blake224, blake256, blake384, blake512,
7  blake2b, blake2s, blake3
8]) {
9  const arr = Uint8Array.from([0x10, 0x20, 0x30]);
10  const a = hash(arr);
11  const b = hash.create().update(arr).digest();
12}
13
14// blake2 advanced usage
15const ab = Uint8Array.from([0x01]);
16blake2s(ab);
17blake2s(ab, { key: new Uint8Array(32) });
18blake2s(ab, { personalization: 'pers1234' });
19blake2s(ab, { salt: 'salt1234' });
20blake2b(ab);
21blake2b(ab, { key: new Uint8Array(64) });
22blake2b(ab, { personalization: 'pers1234pers1234' });
23blake2b(ab, { salt: 'salt1234salt1234' });
24
25// blake3 advanced usage
26blake3(ab);
27blake3(ab, { dkLen: 256 });
28blake3(ab, { key: new Uint8Array(32) });
29blake3(ab, { context: 'application-name' });

• Blake1 is legacy hash, one of SHA3 proposals. It is rarely used anywhere. See pdf.
• Blake2 is popular fast hash. blake2b focuses on 64-bit platforms while blake2s is for 8-bit to 32-bit ones. See RFC 7693, Website
• Blake3 is faster, reduced-round blake2. See Website & specs

legacy: sha1, md5, ripemd160

SHA1 (RFC 3174), MD5 (RFC 1321) and RIPEMD160 (RFC 2286) legacy, weak hash functions. Don't use them in a new protocol. What "weak" means:

• Collisions can be made with 2^18 effort in MD5, 2^60 in SHA1, 2^80 in RIPEMD160.
• No practical pre-image attacks (only theoretical, 2^123.4)
• HMAC seems kinda ok: https://datatracker.ietf.org/doc/html/rfc6151

1import { md5, ripemd160, sha1 } from '@noble/hashes/legacy.js';
2for (let hash of [md5, ripemd160, sha1]) {
3  const arr = Uint8Array.from([0x10, 0x20, 0x30]);
4  const a = hash(arr);
5  const b = hash.create().update(arr).digest();
6}

hmac

1import { hmac } from '@noble/hashes/hmac.js';
2import { sha256 } from '@noble/hashes/sha2.js';
3const key = new Uint8Array(32).fill(1);
4const msg = new Uint8Array(32).fill(2);
5const mac1 = hmac(sha256, key, msg);
6const mac2 = hmac.create(sha256, key).update(msg).digest();

Matches RFC 2104.

hkdf

1import { hkdf } from '@noble/hashes/hkdf.js';
2import { randomBytes } from '@noble/hashes/utils.js';
3import { sha256 } from '@noble/hashes/sha2.js';
4const inputKey = randomBytes(32);
5const salt = randomBytes(32);
6const info = 'application-key';
7const hk1 = hkdf(sha256, inputKey, salt, info, 32);
8
9// == same as
10import { extract, expand } from '@noble/hashes/hkdf.js';
11import { sha256 } from '@noble/hashes/sha2.js';
12const prk = extract(sha256, inputKey, salt);
13const hk2 = expand(sha256, prk, info, 32);

Matches RFC 5869.

pbkdf2

1import { pbkdf2, pbkdf2Async } from '@noble/hashes/pbkdf2.js';
2import { sha256 } from '@noble/hashes/sha2.js';
3const pbkey1 = pbkdf2(sha256, 'password', 'salt', { c: 32, dkLen: 32 });
4const pbkey2 = await pbkdf2Async(sha256, 'password', 'salt', { c: 32, dkLen: 32 });
5const pbkey3 = await pbkdf2Async(sha256, Uint8Array.from([1, 2, 3]), Uint8Array.from([4, 5, 6]), {
6  c: 32,
7  dkLen: 32,
8});

Matches RFC 2898.

scrypt

1import { scrypt, scryptAsync } from '@noble/hashes/scrypt.js';
2const scr1 = scrypt('password', 'salt', { N: 2 ** 16, r: 8, p: 1, dkLen: 32 });
3const scr2 = await scryptAsync('password', 'salt', { N: 2 ** 16, r: 8, p: 1, dkLen: 32 });
4const scr3 = await scryptAsync(Uint8Array.from([1, 2, 3]), Uint8Array.from([4, 5, 6]), {
5  N: 2 ** 17,
6  r: 8,
7  p: 1,
8  dkLen: 32,
9  onProgress(percentage) {
10    console.log('progress', percentage);
11  },
12  maxmem: 2 ** 32 + 128 * 8 * 1, // N * r * p * 128 + (128*r*p)
13});

Conforms to RFC 7914, Website

• N, r, p are work factors. To understand them, see the blog post. r: 8, p: 1 are common. JS doesn't support parallelization, making increasing p meaningless.
• dkLen is the length of output bytes e.g. 32 or 64
• onProgress can be used with async version of the function to report progress to a user.
• maxmem prevents DoS and is limited to 1GB + 1KB (2**30 + 2**10), but can be adjusted using formula: N * r * p * 128 + (128 * r * p)

Time it takes to derive Scrypt key under different values of N (2**N) on Apple M4 (mobile phones can be 1x-4x slower):

N pow	Time	RAM
16	0.1s	64MB
17	0.2s	128MB
18	0.4s	256MB
19	0.8s	512MB
20	1.5s	1GB
21	3.1s	2GB
22	6.2s	4GB
23	13s	8GB
24	27s	16GB

[!NOTE] We support N larger than 2**20 where available, however, not all JS engines support >= 2GB ArrayBuffer-s. When using such N, you'll need to manually adjust maxmem, using formula above. Other JS implementations don't support large N-s.

argon2

1import { argon2d, argon2i, argon2id } from '@noble/hashes/argon2.js';
2const arg1 = argon2id('password', 'saltsalt', { t: 2, m: 65536, p: 1, maxmem: 2 ** 32 - 1 });

Argon2 RFC 9106 implementation.

[!WARNING] Argon2 can't be fast in JS, because there is no fast Uint64Array. It is suggested to use Scrypt instead. Being 5x slower than native code means brute-forcing attackers have bigger advantage.

utils

1import { bytesToHex as toHex, randomBytes } from '@noble/hashes/utils';
2console.log(toHex(randomBytes(32)));

• bytesToHex will convert Uint8Array to a hex string
• randomBytes(bytes) will produce cryptographically secure random Uint8Array of length bytes

Security

The library has been independently audited:

• at version 1.0.0, in Jan 2022, by Cure53
  • PDFs: website, in-repo
  • Changes since audit.
  • Scope: everything, besides blake3, sha3-addons, sha1 and argon2, which have not been audited
  • The audit has been funded by Ethereum Foundation with help of Nomic Labs

It is tested against property-based, cross-library and Wycheproof vectors, and is being fuzzed in the separate repo.

If you see anything unusual: investigate and report.

Constant-timeness

We're targetting algorithmic constant time. JIT-compiler and Garbage Collector make "constant time" extremely hard to achieve timing attack resistance in a scripting language. Which means any other JS library can't have constant-timeness. Even statically typed Rust, a language without GC, makes it harder to achieve constant-time for some cases. If your goal is absolute security, don't use any JS lib — including bindings to native ones. Use low-level libraries & languages.

Memory dumping

The library shares state buffers between hash function calls. The buffers are zeroed-out after each call. However, if an attacker can read application memory, you are doomed in any case:

• At some point, input will be a string and strings are immutable in JS: there is no way to overwrite them with zeros. For example: deriving key from scrypt(password, salt) where password and salt are strings
• Input from a file will stay in file buffers
• Input / output will be re-used multiple times in application which means it could stay in memory
• await anything() will always write all internal variables (including numbers) to memory. With async functions / Promises there are no guarantees when the code chunk would be executed. Which means attacker can have plenty of time to read data from memory
• There is no way to guarantee anything about zeroing sensitive data without complex tests-suite which will dump process memory and verify that there is no sensitive data left. For JS it means testing all browsers (incl. mobile), which is complex. And of course it will be useless without using the same test-suite in the actual application that consumes the library

Supply chain security

• Commits are signed with PGP keys, to prevent forgery. Make sure to verify commit signatures
• Releases are transparent and built on GitHub CI. Make sure to verify provenance logs
  • Use GitHub CLI to verify single-file builds: gh attestation verify --owner paulmillr noble-hashes.js
• Rare releasing is followed to ensure less re-audit need for end-users
• Dependencies are minimized and locked-down: any dependency could get hacked and users will be downloading malware with every install.
  • We make sure to use as few dependencies as possible
  • Automatic dep updates are prevented by locking-down version ranges; diffs are checked with npm-diff
• Dev Dependencies are disabled for end-users; they are only used to develop / build the source code

For this package, there are 0 dependencies; and a few dev dependencies:

• micro-bmark, micro-should and jsbt are used for benchmarking / testing / build tooling and developed by the same author
• prettier, fast-check and typescript are used for code quality / test generation / ts compilation. It's hard to audit their source code thoroughly and fully because of their size

Randomness

We're deferring to built-in crypto.getRandomValues which is considered cryptographically secure (CSPRNG).

In the past, browsers had bugs that made it weak: it may happen again. Implementing a userspace CSPRNG to get resilient to the weakness is even worse: there is no reliable userspace source of quality entropy.

Quantum computers

Cryptographically relevant quantum computer, if built, will allow to utilize Grover's algorithm to break hashes in 2^n/2 operations, instead of 2^n.

This means SHA256 should be replaced with SHA512, SHA3-256 with SHA3-512, SHAKE128 with SHAKE256 etc.

Australian ASD prohibits SHA256 and similar hashes after 2030.

Speed

1npm run bench:install && npm run bench

Benchmarks measured on Apple M4.

# 32B
sha256 x 1,968,503 ops/sec @ 508ns/op
sha512 x 740,740 ops/sec @ 1μs/op
sha3_256 x 287,686 ops/sec @ 3μs/op
sha3_512 x 288,267 ops/sec @ 3μs/op
k12 x 476,190 ops/sec @ 2μs/op
m14 x 423,190 ops/sec @ 2μs/op
blake2b x 464,252 ops/sec @ 2μs/op
blake2s x 766,871 ops/sec @ 1μs/op
blake3 x 879,507 ops/sec @ 1μs/op

# 1MB
sha256 x 331 ops/sec @ 3ms/op
sha512 x 129 ops/sec @ 7ms/op
sha3_256 x 38 ops/sec @ 25ms/op
sha3_512 x 20 ops/sec @ 47ms/op
k12 x 88 ops/sec @ 11ms/op
m14 x 62 ops/sec @ 15ms/op
blake2b x 69 ops/sec @ 14ms/op
blake2s x 57 ops/sec @ 17ms/op
blake3 x 72 ops/sec @ 13ms/op

# MAC
hmac(sha256) x 599,880 ops/sec @ 1μs/op
hmac(sha512) x 197,122 ops/sec @ 5μs/op
kmac256 x 87,981 ops/sec @ 11μs/op
blake3(key) x 796,812 ops/sec @ 1μs/op

# KDF
hkdf(sha256) x 259,942 ops/sec @ 3μs/op
blake3(context) x 424,808 ops/sec @ 2μs/op
pbkdf2(sha256, c: 2 ** 18) x 5 ops/sec @ 197ms/op
pbkdf2(sha512, c: 2 ** 18) x 1 ops/sec @ 630ms/op
scrypt(n: 2 ** 18, r: 8, p: 1) x 2 ops/sec @ 400ms/op
argon2id(t: 1, m: 256MB) 2881ms

Compare to native node.js implementation that uses C bindings instead of pure-js code:

# native (node) 32B
sha256 x 2,267,573 ops/sec
sha512 x 983,284 ops/sec
sha3_256 x 1,522,070 ops/sec
blake2b x 1,512,859 ops/sec
blake2s x 1,821,493 ops/sec
hmac(sha256) x 1,085,776 ops/sec
hkdf(sha256) x 312,109 ops/sec
# native (node) KDF
pbkdf2(sha256, c: 2 ** 18) x 5 ops/sec @ 197ms/op
pbkdf2(sha512, c: 2 ** 18) x 1 ops/sec @ 630ms/op
scrypt(n: 2 ** 18, r: 8, p: 1) x 2 ops/sec @ 378ms/op

It is possible to make this library 4x+ faster by doing code generation of full loop unrolls. We've decided against it. Reasons:

• the library must be auditable, with minimum amount of code, and zero dependencies
• most method invocations with the lib are going to be something like hashing 32b to 64kb of data
• hashing big inputs is 10x faster with low-level languages, which means you should probably pick 'em instead

The current performance is good enough when compared to other projects; SHA256 takes only 900 nanoseconds to run.

Contributing & testing

test/misc directory contains implementations of loop unrolling and md5.

• npm install && npm run build && npm test will build the code and run tests.
• npm run lint / npm run format will run linter / fix linter issues.
• npm run bench will run benchmarks, which may need their deps first (npm run bench:install)
• npm run build:release will build single file
• There is additional 20-min DoS test npm run test:dos and 2-hour "big" multicore test npm run test:big. See our approach to testing

Additional resources:

• NTT hashes are outside of scope of the library. You can view some of them in different repos:
  • Pedersen in micro-zk-proofs
  • Poseidon in noble-curves
• Check out guidelines for coding practices
• See paulmillr.com/noble for useful resources, articles, documentation and demos related to the library.

License

The MIT License (MIT)

Copyright (c) 2022 Paul Miller (https://paulmillr.com)

See LICENSE file.