no-weak-hash-algorithm

Detects usage of weak hash algorithms (MD5, MD4, SHA-1, RIPEMD) in Node.js crypto operations. This rule is part of eslint-plugin-node-security and provides LLM-optimized error messages with fix suggestions.

🚨 Security rule | 💡 Provides suggestions | ⚠️ Set to error in recommended

Quick SummaryCopy Anchor Link

Vulnerability and RiskCopy Anchor Link

Vulnerability: MD5, MD4, SHA-1, and RIPEMD are cryptographically broken hash algorithms. They are vulnerable to collision attacks which allow attackers to create different messages with the same hash.

Risk: Using weak hash algorithms for security-sensitive operations (password hashing, digital signatures, file integrity) can allow attackers to forge signatures, create colliding files, or perform preimage attacks.

Rule DetailsCopy Anchor Link

This rule detects usage of weak hash algorithms in crypto.createHash() calls and suggests secure alternatives like SHA-256, SHA-512, or SHA-3.

Why This MattersCopy Anchor Link

ConfigurationCopy Anchor Link

{
  rules: {
    'node-security/no-weak-hash-algorithm': ['error', {
      additionalWeakAlgorithms: ['whirlpool'],
      allowInTests: false
    }]
  }
}

ExamplesCopy Anchor Link

❌ IncorrectCopy Anchor Link

import crypto from 'crypto';

// MD5 - completely broken
const passwordHash = crypto.createHash('md5').update(data).digest('hex');

// SHA-1 - collision attacks demonstrated
const signatureHash = crypto.createHash('sha1').update(data).digest('hex');

// MD4 - severely broken
const tokenHash = crypto.createHash('md4').update(data).digest('hex');

// RIPEMD - deprecated
const integrityHash = crypto.createHash('ripemd160').update(data).digest('hex');

These examples name a security use (password, signature, token, integrity), which is what makes them fire under the default options. The block previously used neutral names like hash and sha1Hash and therefore reported nothing — see Unclassified hashes below for why, and set reportUnclassifiedHashes: true to also flag weak hashes whose name gives no purpose. Names listed in nonCryptographicNames stay exempt either way.

✅ CorrectCopy Anchor Link

import crypto from 'crypto';

// SHA-256 - recommended for most use cases
const hash = crypto.createHash('sha256').update(data).digest('hex');

// SHA-512 - stronger, use for high-security needs
const sha512Hash = crypto.createHash('sha512').update(data).digest('hex');

// SHA-3 - newest, NIST-approved
const sha3Hash = crypto.createHash('sha3-256').update(data).digest('hex');

Security ImpactCopy Anchor Link

Migration GuideCopy Anchor Link

Phase 1: DiscoveryCopy Anchor Link

{
  rules: {
    'node-security/no-weak-hash-algorithm': 'warn'
  }
}

Phase 2: ReplacementCopy Anchor Link

// Replace MD5/SHA-1 with SHA-256
crypto.createHash('md5'); // ❌ Before
crypto.createHash('sha256'); // ✅ After

crypto.createHash('sha1'); // ❌ Before
crypto.createHash('sha256'); // ✅ After

Phase 3: EnforcementCopy Anchor Link

{
  rules: {
    'node-security/no-weak-hash-algorithm': 'error'
  }
}

Related RulesCopy Anchor Link

• no-sha1-hash - Specific SHA-1 detection for crypto-hash package
• no-weak-cipher-algorithm - Detect weak encryption algorithms
• prefer-native-crypto - Prefer Node.js native crypto

Known False NegativesCopy Anchor Link

The following patterns are not detected due to static analysis limitations:

Dynamic Algorithm NamesCopy Anchor Link

Why: Dynamic strings cannot be analyzed statically.

// ❌ NOT DETECTED - Dynamic algorithm
const algorithm = getAlgorithm();
crypto.createHash(algorithm);

Mitigation: Use constants for algorithm names.

Variable ReassignmentCopy Anchor Link

Why: Cross-function data flow is not tracked.

// ❌ NOT DETECTED - Variable algorithm
let algo = 'md5';
crypto.createHash(algo);

Mitigation: Apply linting at integration points.

Unclassified hashesCopy Anchor Link

Why: This is the largest false-negative class, and it is a deliberate trade rather than an analysis limit. By default (reportUnclassifiedHashes: false) a weak hash is reported only when a securityUseNames word is visible at the site. A digest whose surrounding identifiers name no security use is not reported, however the value is then used:

// ❌ NOT DETECTED - no securityUseNames word at the site
const sha1 = (buf: Buffer) => createHash('sha1').update(buf).digest('hex');
const shasum = sha1(archive);
if (published !== shasum) throw new Error('tarball integrity mismatch');

This is a real supply-chain integrity check on a downloaded archive, and it is silent. Renaming shasum to integrity makes the identical code report CRITICAL, because integrity IS in securityUseNames — so the discriminator is the vocabulary that happens to appear, not the security semantics.

Mitigation: set reportUnclassifiedHashes: true, or name the binding after what it protects so the classifier can see it.

Further ReadingCopy Anchor Link

• NIST Hash Function Guidelines - NIST recommendations
• CWE-327: Broken Crypto Algorithm - Official CWE entry
• Node.js Crypto Documentation - Node.js crypto module

Non-cryptographic hashingCopy Anchor Link

Not every hash is a security control. redis/ioredis computes the EVALSHA script identifier the Redis wire protocol mandates:

this.sha = createHash('sha1').update(lua).digest('hex');

SHA-1 is genuinely used here, but the algorithm is not the author's choice and breaking its collision resistance buys nothing — the value indexes a script the server already holds. The same shape covers HTTP ETags, content-addressed caches, and cache busting.

A hash assigned to a name in nonCryptographicNames is not reported. The test is where the value lands, not which API produced it: the rule walks out through the .update(...).digest(...) receiver chain and reads the assignment target. A hash that is returned, passed as an argument, compared, or stored under a computed key is not exempted by this rule — so renaming a variable to sha cannot silence a hash the classifier had already decided to report.

That is a statement about the nonCryptographicNames exemption only, and it is not a promise that such a hash reports. Under the default reportUnclassifiedHashes: false, a digest with no securityUseNames word at the site is never classified as a security control in the first place, so there is nothing for the exemption to override. See Unclassified hashes.

Set nonCryptographicNames: [] to switch the exemption off.

⚙️ OptionsCopy Anchor Link