1
OAuth 2.0 (RFC 6749) violations
Source: dani-garcia/vaultwarden#5208 · opened by @zacknewman
Sections [10.3]( and [10.4]( of RFC 6749 requires the authorization server to use TLS when exchanging the access and refresh tokens; however Vaultwarden—which acts as both the resource and authorization servers—currently does not force TLS, meaning deployments without TLS violate the RFC. Additionally [Section 6]( requires the refresh token to be uniquely linked to the client; however Vaultwarden does not have a UNIQUE CONSTRAINT defined on devices.refresh_token which means it's possible, albeit unlikely, for multiple clients to have the same refresh token. If adherence to RFC 6749 is not a goal, I do believe a UNIQUE CONSTRAINT should still be defined on devices.refresh_token since it will ensure the token is truly unique. Furthermore [Vaultwarden already (incorrectly) assumes/hopes the refresh token is unique]( and with an actual UNIQUE CONSTRAINT defined—which will be backed by a UNIQUE INDEX with any real-world RDBMS I am aware of—the performance will be even better (an…
No pledges yet. Be the first to back this.
Comments
Similar requests
Feature Request: Support login_hint Parameter for SSO OAuth Authorization Requests
3 votes · 0 comments
Bruteforce security for 2FA
1 vote · 0 comments
Additional configuration for hardening 2FA
2 votes · 0 comments
Feature Request — Yandex 2FA (yaotp scheme)
1 vote · 0 comments
First-class Auth Method abstraction for OAuth/SSO logins (WeChat, Google, etc.)
1 vote · 0 comments
No comments yet.