TOTP Generator

The code is derived, not transmitted

TOTP takes a shared secret and the current Unix time divided by a 30-second step, runs HMAC-SHA1 over them, and truncates the result to six digits. Both sides compute the same value independently — nothing is sent, which is why TOTP works offline and why intercepting the network gives an attacker nothing beyond that 30-second window.

Clock drift is the usual failure

Because the input is time, a device whose clock is wrong produces wrong codes. Servers typically accept the previous and next step to tolerate a minute of skew. A phone that has drifted further fails every attempt, and the fix is enabling automatic network time rather than re-scanning the QR code.

The secret is the account

Anyone holding the base32 secret can generate valid codes forever. It appears in the QR code and in the manual-entry string, so a screenshot of a setup QR sitting in a photo library is a permanent credential. Backup codes deserve the same handling as the secret itself.

What TOTP does and does not stop

It defeats credential stuffing and password reuse completely, since a stolen password alone is useless. It does not stop real-time phishing: a proxy page can capture the code and replay it within its window. Only origin-bound factors such as WebAuthn or passkeys close that, which is why high-value accounts are moving to them.

SHA-1 here is not a weakness

HMAC-SHA1 remains sound despite SHA-1's collision problems, because HMAC does not depend on collision resistance. The algorithm is fixed by RFC 6238 and by every authenticator app, so changing it breaks compatibility for no security gain.