The vulnerability lies in Parse Server's multi-factor authentication (MFA) implementation, specifically concerning the handling of recovery codes. When a user logs in using a recovery code, the server should invalidate that code to prevent its reuse. The analysis of the provided patches shows that the vulnerability is located in the src/Adapters/Auth/mfa.js file. The commits cdc973c3585d8c3dc0f3e57ac64f7b94e641742f and 5cb4df5df7a43ea6a51bbee22b8951bd1f80b28d (for different branches) clearly show the fix. The vulnerable code simply checked for the existence of the recovery code (if (recovery[0] === token || recovery[1] === token)) and, if valid, allowed the login without consuming the code. The patch modifies this logic to find the index of the used recovery code, remove it from the user's authData, and then save the updated user object. The function containing this logic is MFAAdapter.validateAuthData. An attacker who obtains a user's recovery code could exploit this flaw to gain persistent access to the account, as the code would never be invalidated.