SB2026110613 - Multiple vulnerabilities in Doorkeeper
Published: November 6, 2026 Updated: November 6, 2026
Breakdown by Severity
- Low
- Medium
- High
- Critical
Description
This security bulletin contains information about 2 vulnerabilities.
1) Open redirect (CVE-ID: N/A)
CWE-ID: CWE-601 - URL Redirection to Untrusted Site ('Open Redirect')
CVSSv4: 2.1 [CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:A/VC:L/VI:N/VA:N/SC:N/SI:N/SA:N]
The vulnerability allows an unauthenticated remote attacker to redirect an authenticated resource owner to an arbitrary, unregistered destination through Doorkeeper's authorization denial endpoint, facilitating phishing and potential token theft, and under distinct conditions to trigger server errors or invoke credential checking that can advance account-lockout counters.
The vulnerability exists due to the deny action in Doorkeeper::AuthorizationsController processing DELETE /oauth/authorize without validating client_id, redirect_uri, or response_type: the optional client-validation callback applies only to the new action, the destroy action bypasses authorization and client-validity checks, denial copies the raw redirect parameter into a redirectable access_denied response, and authorization-strategy resolution falls back to constantizing an unregistered response type without an allow-list, permitting token-endpoint strategies to be instantiated at the authorization endpoint. An unauthenticated remote attacker can supply an arbitrary redirect_uri, even without client_id or with a registered client and an unregistered destination, to cause a cross-host HTTP 302 redirect carrying the denial error, error description, and attacker-chosen state, an automatically submitted form when response_mode is form_post, or an HTTP 400 JSON response containing the attacker-controlled redirect URI in api_only mode; exploitation against an authenticated victim requires attacker-controlled parameters to reach the deny request, such as through an API-only application forwarding authorization parameters when the victim opens the attacker's link and presses deny, a customized consent form accepting attacker-influenced parameters, or disabled forgery protection combined with SameSite=None session cookies or a same-site attacker origin, whereas default CSRF protection and the default consent form's validated redirect URI prevent straightforward third-party exploitation; successful responses can also include the provider's issuer identifier where supported; when authenticated, the attacker can omit redirect_uri to trigger an unhandled URI::InvalidURIError, HTTP 500 responses, and error-tracker noise, or submit password or client_credentials as response_type to instantiate a token-endpoint strategy lacking a deny method and trigger HTTP 500, with password-strategy construction additionally executing the host application's resource_owner_from_credentials callback even when the password grant is disabled, bypassing rate limits specific to /oauth/token and advancing failed-attempt counters for arbitrary usernames when Devise lockable is used, although the uniform HTTP 500 response does not provide a credential oracle.
2) Improper Authentication (CVE-ID: N/A)
CWE-ID: CWE-287 - Improper Authentication
CVSSv4: 6.9 [CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N]
The vulnerability allows an unauthenticated network attacker to enumerate applications holding active machine-to-machine tokens, revoke their access tokens and grants to deny service to dependent integrations, and access application-management screens exposing client identifiers and, when secrets are stored in plaintext, client secrets.
The vulnerability exists due to Doorkeeper authentication helpers ignoring authenticator return values without halting requests: when authorized-applications routes are mounted and ownerless tokens exist, Doorkeeper::AuthorizedApplicationsController proceeds with a nil resource owner if resource_owner_authenticator is absent and uses its nil-returning default or a custom block returns nil without halting, causing resource-owner queries to select all records with a NULL owner, including under polymorphic_resource_owner, while Doorkeeper::ApplicationsController similarly leaves application-management screens accessible when admin_authenticator returns nil without halting, including from an empty block. An unauthenticated network attacker can send GET requests to /oauth/authorized_applications to obtain application identifiers, names, and creation timestamps and DELETE requests to /oauth/authorized_applications/:id to revoke each application's active client_credentials tokens and grants without credentials, a session, or user interaction, disrupting integrations until new tokens are obtained and controlling when revocation-related logs or alerts fire where configured, with cross-site requests also possible in api_only mode because these routes lack CSRF protection, or access application-management screens under the separate nil-returning administrator-authenticator condition.
Remediation
Install update from vendor's website.