OpenID Connect 1.0

Authelia can act as an OpenID Connect 1.0 Provider as part of an open beta. This section details implementation specifics that can be used for integrating Authelia with an OpenID Connect 1.0 Relying Party, as well as specific documentation for some OpenID Connect 1.0 Relying Party implementations.

See the OpenID Connect 1.0 Provider and OpenID Connect 1.0 Clients configuration guides for information on how to configure the Authelia OpenID Connect 1.0 Provider (note the clients guide is for configuring the registered clients in the provider).

This page is intended as an integration reference point for any implementers who wish to integrate an OpenID Connect 1.0 Relying Party (client application) either as a developer or user of the third party Relying Party.

Audiences#

When it comes to OpenID Connect 1.0 there are effectively two types of audiences. There is the audience embedded in the ID Token which should always include the requesting clients identifier and audience of the Access Token and Refresh Token. The intention of the audience in the ID Token is used to convey which Relying Party or client was the intended audience of the token. In contrast, the audience of the Access Token is used by the Authorization Server or Resource Server to satisfy an internal policy. You could consider the ID Token and it’s audience to be a public facing audience, and the audience of other tokens to be private or have private meaning even when the Access Token is using the JWT Profile for OAuth 2.0 Access Tokens.

It’s also important to note that except RFC9068 there is basically no standardized token format for an Access Token or a Refresh Token. Therefore, there is no way without the use of the Introspection endpoint to determine what audiences these tokens are meant for. It should also be noted that like the scope of a Refresh Token should effectively never change this also applies to the audience of this token.

For these reasons the audience of the Access Token, Refresh Token, and ID Token are effectively completely separate and Authelia treats them in this manner. An ID Token will always and only have the client identifier of the specific client that requested it per specification, the Access Token will always have the granted audience of the Authorization Flow or last successful Refresh Flow, and the Refresh Token will always have the granted audience of the Authorization Flow.

For more information about the opaque Access Token default see Why isn’t the Access Token a JSON Web Token? (Frequently Asked Questions).

Scope Definitions#

The following scope definitions describe each scope supported and the associated effects including the individual claims returned by granting this scope. By default, we do not issue any claims which reveal the users identity which allows administrators semi-granular control over which claims the client is entitled to.

openid#

This is the default scope for OpenID Connect 1.0. This field is forced on every client by the configuration validation that Authelia does.

offline_access#

This scope is a special scope designed to allow applications to obtain a Refresh Token which allows extended access to an application on behalf of a user. A Refresh Token is a special Access Token that allows refreshing previously issued token credentials, effectively it allows the Relying Party to obtain new tokens periodically.

As per OpenID Connect 1.0 Section 11 Offline Access can only be granted during the Authorization Code Flow or a Hybrid Flow. The Refresh Token will only ever be returned at the Token Endpoint when the client is exchanging their OAuth 2.0 Authorization Code.

Generally unless an application supports this and actively requests this scope they should not be granted this scope via the client configuration.

It is also important to note that we treat a Refresh Token as single use and reissue a new Refresh Token during the refresh flow.

groups#

This scope includes the groups the authentication backend reports the user is a member of in the Claims of the ID Token.

email#

This scope includes the email information the authentication backend reports about the user in the Claims of the ID Token.

profile#

This scope includes the profile information the authentication backend reports about the user in the Claims of the ID Token.

Special Scopes#

The following scopes represent special permissions granted to a specific token.

authelia.bearer.authz#

This scope allows the granted access token to be utilized with the bearer authorization scheme on endpoints protected via Authelia.

The specifics about this scope are discussed in the OAuth 2.0 Bearer Token Usage for Authorization Endpoints guide.

Signing and Encryption Algorithms#

OpenID Connect 1.0 and OAuth 2.0 support a wide variety of signature and encryption algorithms. Authelia supports a subset of these.

Response Object#

Authelia’s response objects can have the following signature algorithms:

Request Object#

Authelia accepts a wide variety of request object types. The below table describes these request objects.

Parameters#

The following section describes advanced parameters which can be used in various endpoints as well as their related configuration options.

Response Types#

The following describes the supported response types. See the OAuth 2.0 Multiple Response Type Encoding Practices for more technical information. The default response modes column indicates which response modes are allowed by default on clients configured with this flow type value. The value field is both the required value for the response_type parameter in the authorization request and the response_types client configuration option.

Response Modes#

The following describes the supported response modes. See the OAuth 2.0 Multiple Response Type Encoding Practices for more technical information. The default response modes of a client is based on the Response Types configuration. The value field is both the required value for the response_mode parameter in the authorization request and the response_modes client configuration option.

Grant Types#

The following describes the various OAuth 2.0 and OpenID Connect 1.0 grant types and their support level. The value field is both the required value for the grant_type parameter in the access / token request and the grant_types client configuration option.

Client Authentication Method#

The following describes the supported client authentication methods. See the OpenID Connect 1.0 Client Authentication specification and the OAuth 2.0 - Client Types specification for more information. The value field is the valid values for the token_endpoint_auth_method client configuration option.

Client Assertion Audience#

The client authentication methods which use the JWT Bearer Client Assertions such as client_secret_jwt and private_key_jwt require that the JWT contains an audience (i.e. the aud claim) which exactly matches the full URL for the token endpoint and it must be lowercase.

Per the RFC7523 Section 3: JWT Format and Processing Requirements this claim must be compared using RFC3987 Section 6.2.1: Simple String Comparison and to assist with making this predictable for implementers we ensure the comparison is done against the lowercase form of this URL.

Authentication Method References#

Authelia currently supports adding the amr Claim to the ID Token utilizing the RFC8176 Authentication Method Reference values.

The values this Claim has, are not strictly defined by the OpenID Connect 1.0 specification. As such, some backends may expect a specification other than RFC8176 for this purpose. If you have such an application and wish for us to support it then you’re encouraged to create a feature request.

Below is a list of the potential values we place in the Claim and their meaning:

Introspection Signing Algorithm#

The following table describes the response from the Introspection endpoint depending on the introspection_signing_alg.

When responding with the Signed JSON Web Token the JSON Web Token typ header has the value of token-introspection+jwt.

User Information Signing Algorithm#

The following table describes the response from the UserInfo Endpoint depending on the userinfo_signed_response_alg.

Endpoint Implementations#

The following section documents the endpoints we implement and their respective paths. This information can traditionally be discovered by Relying Parties that utilize OpenID Connect Discovery 1.0, however this information may be useful for clients which do not implement this.

The endpoints can be discovered easily by visiting the Discovery and Metadata endpoints. It is recommended regardless of your version of Authelia that you utilize this version as it will always produce the correct endpoint URLs. The paths for the Discovery/Metadata endpoints are part of IANA’s well known registration but are also documented in a table below.

These tables document the endpoints we currently support and their paths in the most recent version of Authelia. The paths are appended to the end of the primary URL used to access Authelia. The tables use the url https://auth.example.com as an example of the Authelia root URL which is also the OpenID Connect 1.0 Issuer.

Well Known Discovery Endpoints#

These endpoints can be utilized to discover other endpoints and metadata about the Authelia OP.

Discoverable Endpoints#

These endpoints implement OpenID Connect 1.0 Provider specifications.

Security#

The following information covers some security topics some users may wish to be familiar with. All of these elements offer hardening to the flows in differing ways (i.e. some validate the authorization server and some validate the client / Relying Party) which are not essential but recommended.

Pushed Authorization Requests Endpoint#

The Pushed Authorization Requests endpoint is discussed in depth in RFC9126 as well as in the OAuth 2.0 Pushed Authorization Requests documentation.

Essentially it’s a special endpoint that takes the same parameters as the Authorization Endpoint (including Proof Key Code Exchange) with a few caveats:

1. The same Client Authentication mechanism required by the Token Endpoint MUST be used.
2. The request MUST use the HTTP POST method.
3. The request MUST use the application/x-www-form-urlencoded content type (i.e. the parameters MUST be in the body, not the URI).
4. The request MUST occur over the back-channel.

The response of this endpoint is JSON encoded with two key-value pairs:

• request_uri
• expires_in

The expires_in indicates how long the request_uri is valid for. The request_uri is used as a parameter to the Authorization Endpoint instead of the standard parameters (as the request_uri parameter).

The advantages of this approach are as follows:

1. Pushed Authorization Requests cannot be created or influenced by any party other than the Relying Party (client).
2. Since you can force all Authorization requests to be initiated via Pushed Authorization Requests you drastically improve the authorization flows resistance to phishing attacks (this can be done globally or on a per-client basis).
3. Since the Pushed Authorization Requests endpoint requires all of the same Client Authentication mechanisms as the Token Endpoint:
   1. Clients using the confidential Client Type can’t have Pushed Authorization Requests generated by parties who do not have the credentials.
   2. Clients using the public Client Type and utilizing Proof Key Code Exchange never transmit the verifier over any front-channel making even the plain challenge method relatively secure.

OAuth 2.0 Authorization Server Issuer Identification#

The RFC9207: OAuth 2.0 Authorization Server Issuer Identification implementation allows Relying Parties to validate the Authorization Response was returned by the expected issuer by ensuring the response includes the exact issuer in the response. This is an additional check in addition to the state parameter.

This validation is not supported by many clients, but it should be utilized if it is supported.

JWT Secured Authorization Response Mode (JARM)#

The JWT Secured Authorization Response Mode for OAuth 2.0 (JARM) implementation similar to OAuth 2.0 Authorization Server Issuer Identification allows a Relying Party to ensure the Authorization Response was returned by the expected issuer and also ensures the response was not tampered with or forged as it is cryptographically signed.

This response mode is not supported by many clients, but we recommend it is used if it’s supported.

Proof Key Code Exchange#

The Proof Key Code Exchange mechanism is discussed in depth in RFC7636 as well as in the OAuth 2.0 Proof Key Code Exchange documentation.

Essentially a random opaque value is generated by the Relying Party and optionally (but recommended) passed through a SHA256 hash. The original value is saved by the Relying Party, and the hashed value is sent in the Authorization request in the code_verifier parameter with the code_challenge_method set to S256 (or plain using a bad practice of not hashing the opaque value).

When the Relying Party requests the token from the Token Endpoint, they must include the code_verifier parameter again (in the body), but this time they send the value without it being hashed.

The advantages of this approach are as follows:

1. Provided the value was hashed it’s certain that the Relying Party which generated the authorization request is the same party as the one requesting the token or is permitted by the Relying Party to make this request.
2. Even when using the public Client Type there is a form of authentication on the Token Endpoint.