Affected versions of csrf-lite are vulnerable to timing attacks as a result of testing CSRF tokens via a fail-early comparison instead of a constant-time comparison.

Timing attacks remove the exponential increase in entropy gained from increased secret length, by providing per-character feedback on the correctness of a guess via miniscule timing differences.

Under favorable network conditions, an attacker can exploit this to guess the secret in no more than (16*18)288 guesses, instead of the 16^18 guesses required were the timing attack not present.

Recommendation

Update to version 0.1.2 or later.

References

• https://nvd.nist.gov/vuln/detail/CVE-2016-10535
• isaacs/csrf-lite#1
• GHSA-hjhr-r3gq-qvp6
• https://www.npmjs.com/advisories/94