/

CVE-2026-43914

Severity:

HIGH RISK

(7.3)

Authentication Rate Limiting Bypass in Vaultwarden

CVE-2026-43914

Analyzed

Vulnerability type

Category:

Authentication Rate Limiting Bypass

CWE:

CWE-307

Impact:

Account Compromise via Brute-Force Password Guessing

Severity:

HIGH RISK

(7.3)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

Vaultwarden contains an authentication rate-limiting bypass where the email-based 2FA helper endpoint can be abused as a password guessing oracle.

By probing the unprotected /api/two-factor/send-email-login endpoint, an attacker can brute-force username/password combinations without triggering the normal login brute-force protection, even for users who do not have email 2FA configured.

Affected context

This issue affects Vaultwarden deployments that:

  • Expose the /api/two-factor/send-email-login endpoint to the network (e.g., Internet-facing instances)
  • Have email-based two-factor authentication enabled at the server level
  • Rely on Vaultwarden's built-in brute-force protection to limit password guessing attempts

The vulnerability arises because the 2FA email sending endpoint effectively validates username/password pairs without being subject to the same rate limiting as the primary login endpoint.

Exploit preconditions

  • Attack surface: Network-accessible Vaultwarden API, specifically /api/two-factor/send-email-login
  • Required access: No prior authentication; attacker needs only network access to the Vaultwarden instance
  • Trigger point: Repeated invocation of the email 2FA endpoint with candidate username/password pairs
  • Why exploitation is possible: The endpoint's behavior depends on whether the provided credentials are valid, and it is not protected by the same brute-force rate limiting as the standard login flow. This creates an oracle that reveals when a correct username/password combination has been found.

Why this happens

The underlying issue is an inconsistent application of authentication rate limiting across related authentication flows.

While the primary login endpoint enforces brute-force protections, the auxiliary email 2FA endpoint can be called with credentials and responds in a way that distinguishes valid from invalid username/password combinations.

This maps to CWE-307: Improper Restriction of Excessive Authentication Attempts, where secondary or supporting authentication endpoints are not subject to the same rate-limiting and lockout policies as the main login mechanism.

Potential impact

  • Attacker can systematically brute-force passwords for arbitrary accounts without triggering standard login rate limits
  • Unauthorized access to user vaults if weak or reused passwords are successfully guessed
  • Compromise of sensitive stored data (e.g., credentials, secrets, personal information)
  • Potential pivot into other systems and services via stolen credentials
  • Erosion of user trust due to perceived bypass of advertised security protections

General mitigation guidance

Effective mitigation centers on aligning all authentication-related endpoints with consistent rate-limiting and abuse-prevention policies.

Common defensive patterns include:

  • Applying the same brute-force protections (rate limits, lockouts, backoff) to every endpoint that can validate or infer credential correctness, including 2FA helpers and recovery flows
  • Designing 2FA-related APIs so they do not internally re-check username/password combinations, or ensuring such checks are handled by a shared component that already enforces throttling
  • Normalizing responses and timing so auxiliary endpoints do not act as credential validity oracles
  • Introducing centralized authentication middleware that enforces rate limiting and anomaly detection uniformly across login, 2FA, and account recovery routes
  • Encouraging and supporting strong, unique passwords and optional additional factors (e.g., hardware or app-based 2FA) to reduce the effectiveness of brute-force attacks

How teams detect issues like this

Organizations can detect and monitor for this kind of weakness and abuse by:

  • Reviewing authentication code paths to ensure that all endpoints which accept or validate credentials share common rate-limiting and lockout logic
  • Using static analysis or code search to identify routes that:
  • Accept username and password parameters
  • Branch behavior based on credential validity
  • Bypass centralized authentication middleware
  • Performing security tests that attempt high-volume login attempts across both primary and secondary authentication endpoints, verifying that rate limiting is consistently enforced
  • Monitoring logs and metrics for:
  • High-frequency requests to the email 2FA endpoint
  • Repeated requests with varying passwords for the same username
  • Patterns of distributed or scripted login attempts that target non-primary endpoints
  • Employing automated API security testing tools that can simulate abusive authentication flows and identify endpoints acting as oracles

Sources

SourceLink
github.comgithub.com/dani-garcia/vaultwarden/pull/6867
github.comgithub.com/dani-garcia/vaultwarden/releases/tag/1.35.4
github.comgithub.com/dani-garcia/vaultwarden/security/advisories/GHSA-c5rv-q295-7w4g

Frequently asked

questions

Yes. An attacker with network access to the Vaultwarden instance can exploit this vulnerability remotely by sending crafted requests to the 2FA email endpoint.

No existing account access is required. The attacker only needs a target username (or a way to enumerate usernames) and network access to the affected endpoint.

The safest pattern is to centralize authentication and rate limiting so that every endpoint capable of validating credentials or indicating credential correctness is protected by the same throttling, lockout, and anomaly detection logic. Auxiliary endpoints should either delegate to this common mechanism or avoid credential re-validation entirely.

When designing authentication and 2FA flows, treat all related endpoints as part of a single security boundary. Establish shared middleware or services for credential validation, rate limiting, and response normalization, and ensure threat modeling explicitly considers whether any endpoint could act as a credential oracle or bypass established protections.

Find out what's already
exploitable in your codebase.