Internet-Draft | Grant Negotiation and Authorization Prot | October 2020 |
Richer | Expires 12 April 2021 | [Page] |
This document defines a mechanism for delegating authorization to a piece of software, and conveying that delegation to the software. This delegation can include access to a set of APIs as well as information passed directly to the software.¶
This document has been prepared by the GNAP working group design team of Kathleen Moriarty, Fabien Imbault, Dick Hardt, Mike Jones, and Justin Richer. This document is intended as a starting point for the working group and includes decision points for discussion and agreement. Many of the features in this proposed protocol can be accomplished in a number of ways. Where possible, the editor has included notes and discussion from the design team regarding the options as understood.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 12 April 2021.¶
Copyright (c) 2020 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.¶
This protocol allows a piece of software, the resource client, to request delegated authorization to resource servers and direct information. This delegation is facilitated by an authorization server usually on behalf of a resource owner. The requesting party operating the software may interact with the authorization server to authenticate, provide consent, and authorize the request.¶
The process by which the delegation happens is known as a grant, and the GNAP protocol allows for the negotiation of the grant process over time by multiple parties acting in distinct roles.¶
This protocol solves many of the same use cases as OAuth 2.0 [RFC6749], OpenID Connect [OIDC], and the family of protocols that have grown up around that ecosystem. However, GNAP is not an extension of OAuth 2.0 and is not intended to be directly compatible with OAuth 2.0. GNAP seeks to provide functionality and solve use cases that OAuth 2.0 cannot easily or cleanly address. Even so, GNAP and OAuth 2.0 will exist in parallel for many deployments, and considerations have been taken to facilitate the mapping and transition from legacy systems to GNAP. Some examples of these can be found in Appendix D.2.¶
The parties in the GNAP protocol perform actions under different roles. Roles are defined by the actions taken and the expectations leveraged on the role by the overall protocol.¶
The GNAP protocol design does not assume any one deployment architecture, but instead attempts to define roles that can be fulfilled in a number of different ways for different use cases. As long as a given role fulfills all of its obligations and behaviors as defined by the protocol, GNAP does not make additional requirements on its structure or setup.¶
Multiple roles can be fulfilled by the same party, and a given party can switch roles in different instances of the protocol. For example, the RO and RQ in many instances are the same person, where a user is authorizing the RC to act on their own behalf at the RS. In this case, one party fulfills both of the RO and RQ roles, but the roles themselves are still defined separately from each other to allow for other use cases where they are fulfilled by different parties.¶
For another example, in some complex scenarios, an RS receiving requests from one RC can act as an RC for a downstream secondary RS in order to fulfill the original request. In this case, one piece of software is both an RS and an RC from different perspectives, and it fulfills these roles separately as far as the overall protocol is concerned.¶
A single role need not be deployed as a monolithic service. For example, An RC could have components that are installed on the RQ's device as well as a back-end system that it communicates with. If both of these components participate in the delegation protocol, they are both considered part of the RC.¶
For another example, an AS could likewise be built out of many constituent components in a distributed architecture. The component that the RC calls directly could be different from the component that the the RO interacts with to drive consent, since API calls and user interaction have different security considerations in many environments. Furthermore, the AS could need to collect identity claims about the RO from one system that deals with user attributes while generating access tokens at another system that deals with security rights. From the perspective of GNAP, all of these are pieces of the AS and together fulfill the role of the AS as defined by the protocol.¶
[[ Editor's note: The names for the roles are an area of ongoing discussion within the working group, as is the appropriate precision of what activities and expectations a particular role covers. In particular, the AS might be formally decomposed into delegation components, that the client talks to, and interaction components, that the user talks to. Several alternative names have been proposed for different roles and components, including:¶
]]¶
In addition to the roles above, the protocol also involves several elements that are acted upon by the roles throughout the process.¶
[[ Editor's note: What other core elements need an introduction here? These aren't roles to be taken on by different parties, nor are they descriptions of the possible configurations of parties, but these are still important moving parts within the protocol. ]]¶
The GNAP protocol can be used in a variety of ways to allow the core delegation process to take place. Many portions of this process are conditionally present depending on the context of the deployments, and not every step in this overview will happen in all circumstances.¶
Note that a connection between roles in this process does not necessarily indicate that a specific protocol message is sent across the wire between the components fulfilling the roles in question, or that a particular step is required every time. For example, for an RC interested in only getting subject information directly, and not calling an RS, all steps involving the RS below do not apply.¶
In some circumstances, the information needed at a given stage is communicated out-of-band or is pre-configured between the components or entities performing the roles. For example, one entity can fulfil multiple roles, and so explicit communication between the roles is not necessary within the protocol flow.¶
+------------+ +------------+ | Requesting | ~ ~ ~ ~ ~ ~ | Resource | | Party (RQ) | | Owner (RO) | +------------+ +------------+ + + + + (A) (B) + + + + +--------+ + +------------+ |Resource|--------------(1)------+------>| Resource | | Client | + | Server | | (RC) | +---------------+ | (RS) | | |--(2)->| Authorization | | | | |<-(3)--| Server | | | | | | (AS) | | | | |--(4)->| | | | | |<-(5)--| | | | | |--------------(6)------------->| | | | | |<~(7)~~| | | |<-------------(8)------------->| | | |--(9)->| | | | | |<-(10)-| | | | | |--------------(11)------------>| | | | | |<~(12)~| | | |-(13)->| | | | | | | | | | +--------+ +---------------+ +------------+ Legend + + + indicates a possible interaction with a human ----- indicates an interaction between protocol roles ~ ~ ~ indicates a potential equivalence or out-of-band communication between roles¶
The following sections and Appendix C contain specific guidance on how to use the GNAP protocol in different situations and deployments.¶
In this example flow, the RC is a web application that wants access to resources on behalf of the current user, who acts as both the requesting party (RQ) and the resource owner (RO). Since the RC is capable of directing the user to an arbitrary URL and receiving responses from the user's browser, interaction here is handled through front-channel redirects using the user's browser. The RC uses a persistent session with the user to ensure the same user that is starting the interaction is the user that returns from the interaction.¶
+--------+ +--------+ +------+ | RC | | AS | | RO | | | | | | + | | |< (1) + Start Session + + + + + + + + + + + + + + + +| RQ | | | | | |(User)| | |--(2)--- Request Access --------->| | | | | | | | | | | |<-(3)-- Interaction Needed -------| | | | | | | | | | | |+ (4) + Redirect for Interaction + + + + + + + + + > | | | | | | | | | | | |<+ (5) +>| | | | | | AuthN | | | | | | | | | | | |<+ (6) +>| | | | | | AuthZ | | | | | | | | | |< (7) + Redirect for Continuation + + + + + + + + + +| | | | | | +------+ | |--(8)--- Continue Request ------->| | | | | | | |<-(9)----- Grant Access ----------| | | | | | +--------+ +--------+¶
An example set of protocol messages for this method can be found in Appendix C.1.¶
In this example flow, the RC is a device that is capable of presenting a short, human-readable code to the user and directing the user to enter that code at a known URL. The RC is not capable of presenting an arbitrary URL to the user, nor is it capable of accepting incoming HTTP requests from the user's browser. The RC polls the AS while it is waiting for the RO to authorize the request. The user's interaction is assumed to occur on a secondary device. In this example it is assumed that the user is both the RQ and RO, though the user is not assumed to be interacting with the RC through the same web browser used for interaction at the AS.¶
+--------+ +--------+ +------+ | RC | | AS | | RO | | |--(1)--- Request Access --------->| | | + | | | | | | RQ | | |<-(2)-- Interaction Needed -------| | |(User)| | | | | | | | |+ (3) + + Display User Code + + + + + + + + + + + + >| | | | | | | | | | | |<+ (4) + | | | | | |Open URI | | | | | | | | | | | |<+ (5) +>| | | | | | AuthN | | | |--(9)--- Continue Request (A) --->| | | | | | | |<+ (6) +>| | | |<-(10)- Not Yet Granted (Wait) ---| | Code | | | | | | | | | | | |<+ (7) +>| | | | | | AuthZ | | | | | | | | | | | |<+ (8) +>| | | | | |Completed| | | | | | | | | |--(11)-- Continue Request (B) --->| | +------+ | | | | | |<-(12)----- Grant Access ---------| | | | | | +--------+ +--------+¶
An example set of protocol messages for this method can be found in Appendix C.2.¶
In this example flow, the RQ and RO roles are fulfilled by different parties, and the RO does not interact with the RC. The AS reaches out asynchronously to the RO during the request process to gather the RO's authorization for the RC's request. The RC polls the AS while it is waiting for the RO to authorize the request.¶
+--------+ +--------+ +------+ | RC | | AS | | RO | | |--(1)--- Request Access --------->| | | | | | | | | | | |<-(2)-- Not Yet Granted (Wait) ---| | | | | | | |<+ (3) +>| | | | | | AuthN | | | |--(6)--- Continue Request (A) --->| | | | | | | |<+ (4) +>| | | |<-(7)-- Not Yet Granted (Wait) ---| | AuthZ | | | | | | | | | | | |<+ (5) +>| | | | | |Completed| | | | | | | | | |--(8)--- Continue Request (B) --->| | +------+ | | | | | |<-(9)------ Grant Access ---------| | | | | | +--------+ +--------+¶
An example set of protocol messages for this method can be found in Appendix D.1.¶
In this example flow, the AS policy allows the RC to make a call on its own behalf, without the need for a RO to be involved at runtime to approve the decision. Since there is no explicit RO, the RC does not interact with an RO.¶
+--------+ +--------+ | RC | | AS | | |--(1)--- Request Access --------->| | | | | | | |<-(2)---- Grant Access -----------| | | | | | +--------+ +--------+¶
An example set of protocol messages for this method can be found in Appendix D.¶
In this example flow, the RC receives an access token to access a resource server through some valid GNAP process. The RC uses that token at the RS for some time, but eventually the access token expires. The RC then gets a new access token by rotating the expired access token at the AS using the token's management URL.¶
+--------+ +--------+ | RC | | AS | | |--(1)--- Request Access ----------------->| | | | | | | |<-(2)--- Grant Access --------------------| | | | | | | | +--------+ | | | |--(3)--- Access Resource --->| RS | | | | | | | | | | |<-(4)--- Error Response -----| | | | | | +--------+ | | | | | | | |--(5)--- Rotate Token ------------------->| | | | | | | |<-(6)--- Rotated Token -------------------| | | | | | +--------+ +--------+¶
To start a request, the RC sends JSON [RFC8259] document with an object as its root. Each member of the request object represents a different aspect of the RC's request. Each field is described in detail in a section below.¶
Additional members of this request object can be defined by extensions to this protocol as described in Section 2.9¶
A non-normative example of a grant request is below:¶
{ "resources": [ { "type": "photo-api", "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }, "dolphin-metadata" ], "client": { "display": { "name": "My Client Display Name", "uri": "https://example.net/client" }, "key": { "proof": "jwsd", "jwk": { "kty": "RSA", "e": "AQAB", "kid": "xyz-1", "alg": "RS256", "n": "kOB5rR4Jv0GMeL...." } } }, "interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.example.net/return/123455", "nonce": "LKLTI25DK82FX4T4QFZC" } }, "capabilities": ["ext1", "ext2"], "subject": { "sub_ids": ["iss-sub", "email"], "assertions": ["id_token"] } }¶
The request MUST be sent as a JSON object in the body of the HTTP
POST request with Content-Type application/json
,
unless otherwise specified by the signature mechanism.¶
If the RC is requesting one or more access tokens for the
purpose of accessing an API, the RC MUST include a resources
field. This field MUST be an array (for a single access token (Section 2.1.1)) or
an object (for multiple access tokens (Section 2.1.3)), as described in the following
sections.¶
When requesting an access token, the RC MUST send a
resources
field containing a JSON array. The elements of the JSON
array represent rights of access that the RC is requesting in
the access token. The requested access is the sum of all elements
within the array.¶
The RC declares what access it wants to associated with the
resulting access token using objects that describe multiple
dimensions of access. Each object contains a type
property that determines the type of API that the RC is calling.¶
The value of this field is under the control of the AS. This field MUST be compared using an exact byte match of the string value against known types by the AS. The AS MUST ensure that there is no collision between different authorization data types that it supports. The AS MUST NOT do any collation or normalization of data types during comparison. It is RECOMMENDED that designers of general-purpose APIs use a URI for this field to avoid collisions between multiple API types protected by a single AS.¶
While it is expected that many APIs will have its own properties, a set of common properties are defined here. Specific API implementations SHOULD NOT re-use these fields with different semantics or syntax. The available values for these properties are determined by the API being protected at the RS.¶
[[ Editor's note: this will align with OAuth 2 RAR, but the details of exactly how it aligns are TBD. Since RAR needs to work in the confines of OAuth 2, RAR has to define how to interact with "scope", "resource", and other existing OAuth 2 mechanisms that don't exist in GNAP. ]].¶
The following non-normative example shows the use of both common
and API-specific fields as part of two different access type
values.¶
"resources": [ { "type": "photo-api", "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }, { "type": "financial-transaction", "actions": [ "withdraw" ], "identifier": "account-14-32-32-3", "currency": "USD" } ]¶
If this request is approved, the resulting access token (Section 3.2.1) will include the sum of both of the requested types of access.¶
Instead of sending an object describing the requested resource (Section 2.1.1), a RC MAY send a string known to the AS or RS representing the access being requested. Each string SHOULD correspond to a specific expanded object representation at the AS.¶
[[ Editor's note: we could describe more about how the expansion would work. For example, expand into an object where the value of the "type" field is the value of the string. Or we could leave it open and flexible, since it's really up to the AS/RS to interpret. ]]¶
"resources": [ "read", "dolphin-metadata", "some other thing" ]¶
This value is opaque to the RC and MAY be any valid JSON string, and therefore could include spaces, unicode characters, and properly escaped string sequences. However, in some situations the value is intended to be seen and understood be the RC developer. In such cases, the API designer choosing any such human-readable strings SHOULD take steps to ensure the string values are not easily confused by a developer¶
This functionality is similar in practice to OAuth 2's scope
parameter [RFC6749], where a single string
represents the set of access rights requested by the RC. As such, the reference
string could contain any valid OAuth 2 scope value as in Appendix D.2. Note that the reference
string here is not bound to the same character restrictions as in OAuth 2's scope
definition.¶
A single "resources" array MAY include both object-type and string-type resource items.¶
"resources": [ { "type": "photo-api", "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }, "read", "dolphin-metadata", { "type": "financial-transaction", "actions": [ "withdraw" ], "identifier": "account-14-32-32-3", "currency": "USD" }, "some other thing" ]¶
[[ Editor's note: passing resource requests by reference really is akin to a "scope", and we have many years of experience showing us that the simplicity of giving a developer a set of strings to send is a simple and powerful pattern. We could always require objects and just use the "type" field as a scope value, but that's a lot of complexity to pay for the simple case. Client developers will always know which kind they need to send, because they're picking from the API's documentation. ]]¶
When requesting multiple access tokens, the resources field is a JSON object. The names of the JSON object fields are token identifiers chosen by the RC, and MAY be any valid string. The values of the JSON object fields are JSON arrays representing a single access token request, as specified in requesting a single access token (Section 2.1.1).¶
The following non-normative example shows a request for two
separate access tokens, token1
and token2
.¶
"resources": { "token1": [ { "type": "photo-api", "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }, "dolphin-metadata" ], "token2": [ { "type": "walrus-access", "actions": [ "foo", "bar" ], "locations": [ "https://resource.other/" ], "datatypes": [ "data", "pictures", "walrus whiskers" ] } ] }¶
Any approved access requests are returned in the multiple access token response (Section 3.2.2) structure using the token identifiers in the request.¶
While the AS is ultimately in control of how tokens are returned and bound to the RC, sometimes the RC has context about what it can support that can affect the AS's response. This specification defines several flags that are passed as resource reference strings (Section 2.1.2).¶
Each flag applies only to the single resource request in which it appears.¶
Support of all flags is optional, such as any other resource reference value.¶
proof
mechanism used by the client with a key
value of true
, indicating
the client's presented key is to be used for binding.
[[ Editor's note: should there be a different flag and mechanism for the client
to explicitly indicate which binding method it wants to use, especially if
the client wants to use a different method at the AS than the RS? ]]¶
The AS MUST respond with any applied flags in the token response (Section 3.2)
resources
section.¶
In this non-normative example, the requested access token is to be bound to the client's key and should be kept during rotation.¶
"resources": [ { "type": "photo-api", "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }, "read", "bind_token", "multi_token" ]¶
Additional flags can be registered in a registry TBD (Section 12).¶
[[ Editor's note: while these reference values are "reserved", the ultimate decider for what a reference means is the AS, which means an AS could arguably decide that one of these values means something else. Also, this kind of reservation potentially steps on API namespaces, which OAuth 2 is careful not to do but common extensions like OIDC do with their own scope definitions. However, in OIDC, several "scope" values have behavior similar to what's defined here, particularly "openid" turns on ID tokens in the response and "offline_access" signals for the return of a refresh token, and these can be used outside of OpenID Connect itself. However, to keep these flags out of the general API namespace, we could use a different syntax for sending them. In particular, they could be defined under a GNAP-specific "type" object, where all the flags are fields on the object.¶
resources: [ { type: "gnap-flags", flag1: true, flag2: false, flag3: true ... }, "reference1", "scope2", ... ]¶
Alternatively, all the flags could be sent in an array separate from the rest of the request.¶
resources: [ "reference1", "scope2", ["flag1", "flag2", "flag3"] ... ]¶
This whole thing might also belong in an extension, as it's advanced behavior signaling for very specific cases. However, it seems other extensions would be likely to extend this kind of thing, like OIDC did with "offline_access". ]]¶
If the RC is requesting information about the RO from
the AS, it sends a subject
field as a JSON object. This object MAY
contain the following fields (or additional fields defined in
a registry TBD (Section 12)).¶
id_token
for an [OIDC] ID Token and saml2
for a SAML 2 assertion. Additional
assertion values are defined by a registry TBD (Section 12).
[[ Editor's note: These values are lifted from [RFC8693]'s "token type
identifiers" list, but is there a better source?]]¶
"subject": { "sub_ids": [ "iss-sub", "email" ], "assertions": [ "id_token", "saml2" ] }¶
The AS can determine the RO's identity and permission for releasing this information through interaction with the RO (Section 4), AS policies, or assertions presented by the RC (Section 2.4). If this is determined positively, the AS MAY return the RO's information in its response (Section 3.4) as requested.¶
Subject identifiers requested by the RC serve only to identify the RO in the context of the AS and can't be used as communication channels by the RC, as discussed in Section 3.4. One method of requesting communication channels and other identity claims are discussed in Section 2.8.¶
The AS SHOULD NOT re-use subject identifiers for multiple different ROs.¶
[[ Editor's Note: What we're really saying here is that "even if the AS gives you an email address to identify the user, that isn't a claim that this is a valid email address for that current user, so don't try to email them." In order to get a workable email address, or anything that you can use to contact them, you'd need a full identity protocol and not just this. Also, subject identifiers are asserted by the AS and therefore naturally scoped to the AS. Would changing the name to "as_sub_ids" or "local_sub_ids" help convey that point? ]]¶
Note: the "sub_ids" and "assertions" request fields are independent of each other, and a returned assertion MAY omit a requested subject identifier.¶
[[ Editor's note: we're potentially conflating these two types in the same structure, so perhaps these should be split. There's also a difference between user information and authentication event information. ]]¶
When sending a non-continuation request to the AS, the RC MUST identify
itself by including the client
field of the request and by signing the
request as described in Section 8. Note that for a
continuation request (Section 5), the RC instance is identified by its
association with the request being continued and so this field is
not sent under those circumstances.¶
When RC information is sent
by value, the client
field of the request consists of a JSON
object with the following fields.¶
"client": { "key": { "proof": "httpsig", "jwk": { "kty": "RSA", "e": "AQAB", "kid": "xyz-1", "alg": "RS256", "n": "kOB5rR4Jv0GMeLaY6_It_r3ORwdf8ci_JtffXyaSx8xY..." }, "cert": "MIIEHDCCAwSgAwIBAgIBATANBgkqhkiG9w0BAQsFA..." }, "class_id": "web-server-1234", "display": { "name": "My Client Display Name", "uri": "https://example.net/client" } }¶
Additional fields are defined in a registry TBD (Section 12).¶
The RC MUST prove possession of any presented key by the proof
mechanism
associated with the key in the request. Proof types
are defined in a registry TBD (Section 12) and an initial set of methods
is described in Section 8.¶
Note that the AS MAY know the RC's public key ahead of time, and the AS MAY apply different policies to the request depending on what has been registered against that key. If the same public key is sent by value on subsequent access requests, the AS SHOULD treat these requests as coming from the same RC software instance for purposes of identification, authentication, and policy application. If the AS does not know the RC's public key ahead of time, the AS MAY accept or reject the request based on AS policy, attestations within the client request, and other mechanisms.¶
[[ Editor's note: additional client attestation frameworks will eventually need to be addressed here. For example, the organization the client represents, or a family of client software deployed in a cluster, or the posture of the device the client is installed on. These all need to be separable from the client's key and potentially the instance identifier. ]]¶
If the RC has an instance identifier that the AS can use to determine
appropriate key information, the RC can send this value in the instance_id
field. The instance identifier MAY be assigned to an RC instance at runtime
through the Section 3.5 or MAY be obtained in another fashion,
such as a static registration process at the AS.¶
"client": { "instance_id": "client-541-ab" }¶
If there are no additional fields to send, the RC MAY send the instance identifier as a direct reference value in lieu of the object.¶
"client": "client-541-ab"¶
When the AS receives a request with an instance identifier, the AS MUST ensure that the key used to sign the request (Section 8) is associated with the instance identifier.¶
If the instance_id
field is sent, it MUST NOT be accompanied by other fields unless such
fields are explicitly marked safe for inclusion alongside the instance
identifier.¶
[[ Editor's note: It seems clear that an instance identifier is mutually exclusive with most of the fields in the request (eg, we don't want an attacker being able to swap out a client's registered key just by accessing the identifier). However, some proposed concepts might fit alongside an instance identifier that change at runtime, such as device posture or another dynamic attestation. Should these be sent in the "client" block alongside the instance identifier, should there be a separate top-level block for runtime attestations, or some other mechanism? ]]¶
If the AS does not recognize the instance identifier, the request MUST be rejected with an error.¶
If the RC instance is identified in this manner, the registered key for the RC MAY be a symmetric key known to the AS. The RC MUST NOT send a symmetric key by value in the request, as doing so would expose the key directly instead of proving possession of it.¶
[[ Editor's note: In many ways, passing an instance identifier is analogous to OAuth 2's "client_id" parameter [RFC6749], especially when coupled with a confidential client's registration and authentication process. See Appendix D.2 for an example. Something like this is required to make things easier for client developers in the common case where the AS already knows the client's key, and to allow symmetric keys. ]]¶
The RC key MUST be a public key in at least one supported format and MUST be applicable to the proofing mechanism used in the request. If the key is sent in multiple formats, all the keys MUST be the same. The key presented in this field MUST be the key used to sign the request.¶
Additional key types are defined in a registry TBD (Section 12).¶
[[ Editor's note: we will eventually want to have fetchable keys, I would guess. Things like DID for key identification are going to be important. ]]¶
This non-normative example shows a single key presented in multiple formats using a single proofing mechanism.¶
"key": { "proof": "jwsd", "jwk": { "kty": "RSA", "e": "AQAB", "kid": "xyz-1", "alg": "RS256", "n": "kOB5rR4Jv0GMeLaY6_It_r3ORwdf8ci_JtffXyaSx8xY..." }, "cert": "MIIEHDCCAwSgAwIBAgIBATANBgkqhkiG9w0BAQsFA..." }¶
Continuation requests (Section 5) MUST use the same key (or its most recent rotation) and proof method as the initial request.¶
If the RC has additional information to display to the RO during any interactions at the AS, it MAY send that information in the "display" field. This field is a JSON object that declares information to present to the RO during any interactive sequences.¶
"display": { "name": "My Client Display Name", "uri": "https://example.net/client" }¶
[[ Editor's note: would we want to support pushing a display logo by value? On the upside it allows for more dynamic detached clients and doesn't require the AS to fetch information. On the downside, this is harder for the AS to enforce a policy about and could lead to potential exploits caused by sending binary image files. ]]¶
Additional display fields are defined by a registry TBD (Section 12).¶
The AS SHOULD use these values during interaction with the RO. The values are for informational purposes only and MUST NOT be taken as authentic proof of the RC's identity or source. The AS MAY restrict display values to specific RC instances, as identified by their keys in Section 2.3.¶
If the presented key is known to the AS and is associated with a single instance of the RC software, the process of presenting a key and proving possession of that key is sufficient to authenticate the RC to the AS. The AS MAY associate policies with the RC software identified by this key, such as limiting which resources can be requested and which interaction methods can be used. For example, only specific RCs with certain known keys might be trusted with access tokens without the AS interacting directly with the RO as in Appendix D.¶
The presentation of a key allows the AS to strongly associate multiple successive requests from the same RC with each other. This is true when the AS knows the key ahead of time and can use the key to authenticate the RC software, but also if the key is ephemeral and created just for this request. As such the AS MAY allow for RCs to make requests with unknown keys. This pattern allows for ephemeral RCs, such as single-page applications, and RCs with many individual instances, such as mobile applications, to generate their own key pairs and use them within the protocol without having to go through a separate registration step. The AS MAY limit which capabilities are made available to RCs with unknown keys. For example, the AS could have a policy saying that only previously-registered RCs can request particular resources, or that all RCs with unknown keys have to be interactively approved by an RO.¶
If the RC knows the identity of the RQ through one or more identifiers or assertions, the RC MAY send that information to the AS in the "user" field. The RC MAY pass this information by value or by reference.¶
id_token
for an [OIDC] ID Token and saml2
for a SAML 2 assertion. Additional
assertion values are defined by a registry TBD (Section 12).
[[ Editor's note: These keys are lifted from [RFC8693]'s "token type
identifiers" list, but is there a better source?
Additionally: should this be an array of objects with internal
typing like the sub_ids? Do we expect more than one assertion per
user anyway? ]]¶
"user": { "sub_ids": [ { "subject_type": "email", "email": "user@example.com" } ], "assertions": { "id_token": "eyj..." } }¶
Subject identifiers are hints to the AS in determining the RO and MUST NOT be taken as declarative statements that a particular RO is present at the RC and acting as the RQ. Assertions SHOULD be validated by the AS. [[ editor's note: is this a MUST? Assertion validation is extremely specific to the kind of assertion in place, what other guidance and requirements can we put in place here? ]]¶
If the identified RQ does not match the RO present at the AS during an interaction step, the AS SHOULD reject the request with an error.¶
[[ Editor's note: we're potentially conflating identification (sub_ids) and provable presence (assertions and a trusted reference handle) in the same structure, so perhaps these should be split. The security parameters are pretty different here. ]]¶
If the AS trusts the RC to present verifiable assertions, the AS MAY decide, based on its policy, to skip interaction with the RO, even if the RC provides one or more interaction modes in its request.¶
User reference identifiers can be dynamically issued by the AS (Section 3.5) to allow the RC to represent the same RQ to the AS over subsequent requests.¶
If the RC has a reference for the RQ at this AS, the RC MAY pass that reference as a string. The format of this string is opaque to the RC.¶
"user": "XUT2MFM1XBIKJKSDU8QM"¶
User reference identifiers are not intended to be human-readable user identifiers or structured assertions. For the RC to send either of these, use the full user request object (Section 2.4) instead.¶
[[ Editor's note: we might be able to fold this function into an unstructured user assertion reference issued by the AS to the RC. We could put it in as an assertion type of "gnap_reference" or something like that. Downside: it's more verbose and potentially confusing to the client developer to have an assertion-like thing that's internal to the AS and not an assertion. ]]¶
If the AS does not recognize the user reference, it MUST return an error.¶
Many times, the AS will require interaction with the RO in order to
approve a requested delegation to the RC for both resources and direct
claim information. Many times the RQ using the RC is the same person as
the RO, and the RC can directly drive interaction with the AS by redirecting
the RQ on the same device, or by launching an application. Other times, the
RC can provide information to start the RO's interaction on a secondary
device, or the RC will wait for the RO to approve the request asynchronously.
The RC could also be signaled that interaction has completed by the AS making
callbacks. To facilitate all of these modes, the RC declares the means that it
can interact using the interact
field.¶
The interact
field is a JSON object with keys that declare
different interaction modes. A RC MUST NOT declare an
interaction mode it does not support.
The RC MAY send multiple modes in the same request.
There is no preference order specified in this request. An AS MAY
respond to any, all, or none of the presented interaction modes (Section 3.3) in a request, depending on
its capabilities and what is allowed to fulfill the request. This specification
defines the following interaction modes:¶
The following sections detail requests for interaction modes. Additional interaction modes are defined in a registry TBD (Section 12).¶
[[ Editor's note: there need to be more examples (Appendix C) that knit together the interaction modes into common flows, like an authz-code equivalent. But it's important for the protocol design that these are separate pieces to allow such knitting to take place. ]]¶
In this non-normative example, the RC is indicating that it can redirect (Section 2.5.1) the RQ to an arbitrary URL and can receive a callback (Section 2.5.3) through a browser request.¶
"interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.example.net/return/123455", "nonce": "LKLTI25DK82FX4T4QFZC" } }¶
In this non-normative example, the RC is indicating that it can display a use code (Section 2.5.4) and direct the RQ to an arbitrary URL of maximum length (Section 2.5.1.1) 255 characters, but it cannot accept a callback.¶
"interact": { "redirect": 255, "user_code": true }¶
If the RC does not provide a suitable interaction mechanism, the AS cannot contact the RO asynchronously, and the AS determines that interaction is required, then the AS SHOULD return an error since the RC will be unable to complete the request without authorization.¶
The AS SHOULD apply suitable timeouts to any interaction mechanisms provided, including user codes and redirection URLs. The RC SHOULD apply suitable timeouts to any callback URLs.¶
If the RC is capable of directing the RQ to a URL defined by the AS at runtime, the RC indicates this by sending the "redirect" field with the boolean value "true". The means by which the RC will activate this URL is out of scope of this specification, but common methods include an HTTP redirect, launching a browser on the RQ's device, providing a scannable image encoding, and printing out a URL to an interactive console.¶
"interact": { "redirect": true }¶
If this interaction mode is supported for this RC and request, the AS returns a redirect interaction response Section 3.3.1.¶
If the RC would prefer to redirect to a shortened URL defined by the AS at runtime, the RC indicates this by sending the "redirect" field with an integer indicating the maximum character length of the returned URL. The AS MAY use this value to decide whether to return a shortened form of the response URL. If the AS cannot shorten its response URL enough to fit in the requested size, the AS SHOULD return an error. [[ Editor's note: Or maybe just ignore this part of the interaction request? ]]¶
"interact": { "redirect": 255 }¶
If this interaction mode is supported for this RC and request, the AS returns a redirect interaction response with short URL Section 3.3.1.¶
If the RC can open a URL associated with an application on the RQ's device, the RC indicates this by sending the "app" field with boolean value "true". The means by which the RC determines the application to open with this URL are out of scope of this specification.¶
"interact": { "app": true }¶
If this interaction mode is supported for this RC and request, the AS returns an app interaction response with an app URL payload Section 3.3.2.¶
[[ Editor's note: this is similar to the "redirect" above today as most apps use captured URLs, but there seems to be a desire for splitting the web-based interaction and app-based interaction into different URIs. There's also the possibility of wanting more in the payload than can be reasonably put into the URL, or at least having separate payloads. ]]¶
If the RC is capable of receiving a message from the AS indicating that the RO has completed their interaction, the RC indicates this by sending the "callback" field. The value of this field is an object containing the following members.¶
redirect
Section 2.5.3.1
and push
Section 2.5.3.2, with other values
defined by a registry TBD (Section 12).¶
sha3
or sha2
. If
absent, the default value is sha3
.
[[ Editor's note: This should
be expandable via a registry of cryptographic options, and it
would be good if we didn't define our own identifiers here.
See also note about cryptographic functions in Section 4.4.3.
]]¶
"interact": { "callback": { "method": "redirect", "uri": "https://client.example.net/return/123455", "nonce": "LKLTI25DK82FX4T4QFZC" } }¶
If this interaction mode is supported for this RC and request, the AS returns a nonce for use in validating the callback response (Section 3.3.3). Requests to the callback URI MUST be processed as described in Section 4.4, and the AS MUST require presentation of an interaction callback reference as described in Section 5.1.¶
[[ Editor's note: There has been some call for a post-interaction redirect that is not tied to the underlying security model - specifically, sending the user over to a client-hosted page with client-specific instructions on how to continue. This would be something hosted externally to the client instance, so the client instance would never see this incoming call. We could accomplish that using this "callback" post-redirect mechanism but with "method": "static" or "nonce": false or some other signal to indicate that the client won't see the incoming request. ]]¶
[[ Editor's note: The callback information could alternatively be combined with other methods like "redirect", essentially putting everything in the "callback" object into the field for the other objects. However, this would require each method to define its own set of rules about how callbacks can be used, and we would want them all to be consistent with each other with clear information about how the AS is supposed to respond to all of these.¶
"interact" { "redirect": { "method": "redirect", "uri": "https://client.example.net/return/123455", "nonce": "LKLTI25DK82FX4T4QFZC" } }¶
So if the object is there, you do the redirect on completion, if the object isn't there (it's a boolean, like today), you don't redirect when you're done. Previous versions of this specification used this structure, but it was abandoned in favor of the current setup to allow for different combinations of user interaction methods at the same time while still keeping a consistent security model. OAuth 2's "grant_type" model has proved to be limiting in unanticipated ways since it requires an entirely new grant type to be invented any time there is a new combination of aspects, or it requires each grant type to have many of the same optionalities. Combining these fields back into one, in this way, would allow a client to declare that it expects a callback in response to one kind of interaction method but not others, and include multiple combinations at once. For example, if a client wants to allow a user to redirect to the AS and back on the same device, or to use a usercode on a secondary device without a callback, and the client wants to offer both modes simultaneously. This could alternately be accomplished by allowing the client to "bundle" interaction parameters together, if desirable - for example, if "interact" were an array, the client would accept any combination represented by one object. This example binds the "callback" only to the first "redirect" method, and second (short) "redirect" and "user_code" method do not use a callback.¶
"interact": [ { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.example.net/return/123455", "nonce": "LKLTI25DK82FX4T4QFZC" } }, { "redirect": 255, "user_code": true } ]¶
It's not clear what a response to such an array would be. Would the AS pick one of these bundles? Would it be allowed to respond to any or all of them? Could an AS use different URIs for each bundle? (This seems likely, at least.) Would there be a security problem if the AS used the same URI for both bundles, since one requires a front channel redirect and the other does not?¶
]]¶
A callback method
value of redirect
indicates that the RC
will expect a call from the RO's browser using the HTTP method
GET as described in Section 4.4.1.¶
"interact": { "callback": { "method": "redirect", "uri": "https://client.example.net/return/123455", "nonce": "LKLTI25DK82FX4T4QFZC" } }¶
Requests to the callback URI MUST be processed by the RC as described in Section 4.4.1.¶
Since the incoming request to the callback URL is from the RO's browser, this method is usually used when the RO and RQ are the same entity. As such, the RC MUST ensure the RQ is present on the request to prevent substitution attacks.¶
A callback method
value of push
indicates that the RC will
expect a call from the AS directly using the HTTP method POST
as described in Section 4.4.2.¶
"interact": { "callback": { "method": "push", "uri": "https://client.example.net/return/123455", "nonce": "LKLTI25DK82FX4T4QFZC" } }¶
Requests to the callback URI MUST be processed by the RC as described in Section 4.4.2.¶
Since the incoming request to the callback URL is from the AS and not from the RO's browser, the RC MUST NOT require the RQ to be present on incoming HTTP the request.¶
[[ Editor's note: This post-interaction method can be used in advanced use cases like asynchronous authorization, or simply to signal the client that it should move to the next part of the protocol, even when there is no user present at the client. As such it can feel a little odd being inside the "interact" block of the protocol, but it does align with the redirect-based "callback" method and it seems they really should be mutually-exclusive. Additionally, should there be a method for simply pushing the updated response directly to the client, instead? ]]¶
If the RC is capable of displaying or otherwise communicating a short, human-entered code to the RO, the RC indicates this by sending the "user_code" field with the boolean value "true". This code is to be entered at a static URL that does not change at runtime, as described in Section 3.3.4.¶
"interact": { "user_code": true }¶
If this interaction mode is supported for this RC and request, the AS returns a user code and interaction URL as specified in Section 4.2.¶
If the RC knows the RQ's locale and language preferences, the
RC can send this information to the AS using the ui_locales
field
with an array of locale strings as defined by [RFC5646].¶
"interact": { "ui_locales": ["en-US", "fr-CA"] }¶
If possible, the AS SHOULD use one of the locales in the array, with preference to the first item in the array supported by the AS. If none of the given locales are supported, the AS MAY use a default locale.¶
Additional interaction modes are defined in a registry TBD (Section 12).¶
[[ Editor's note: we should have guidance in here about how to define other interaction modes. There's already interest in defining message-based protocols like DIDCOMM and challenge-response protocols like FIDO, for example. ]]¶
If the RC supports extension capabilities, it MAY present them to the AS in the "capabilities" field. This field is an array of strings representing specific extensions and capabilities, as defined by a registry TBD (Section 12).¶
"capabilities": ["ext1", "ext2"]¶
If the RC has a reference handle from a previously granted
request, it MAY send that reference in the "existing_grant" field. This
field is a single string consisting of the value
of the access_token
returned in a previous request's continuation response (Section 3.1).¶
"existing_grant": "80UPRY5NM33OMUKMKSKU"¶
The AS MUST dereference the grant associated with the reference and process this request in the context of the referenced one. The AS MUST NOT alter the existing grant associated with the reference.¶
[[ Editor's note: this basic capability is to allow for both step-up authorization and downscoped authorization, but by explicitly creating a new request and not modifying an existing one. What's the best guidance for how an AS should process this? What are the use cases that help differentiate this from modification of an existing request? ]]¶
If the RC and AS both support OpenID Connect's claims query language as defined in [OIDC] Section 5.5,
the RC sends the value of the OpenID Connect claims
authorization request parameter as a JSON object
under the name claims
in the root of the request.¶
"claims": { "id_token" : { "email" : { "essential" : true }, "email_verified" : { "essential" : true } }, "userinfo" : { "name" : { "essential" : true }, "picture" : null } }¶
The contents of the claims
parameter have the same semantics as they do in OpenID Connect's
claims
authorization request parameter,
including all extensions such as [OIDC4IA]. The AS MUST process the claims object in the same
way that it would with an OAuth 2 based authorization request.¶
Note that because this is an independent query object, the claims
value can augment or alter
other portions of the request, namely the resources
and subject
fields. This query language uses
the fields in the top level of the object to indicate the target for any requested claims. For instance, the
userinfo
target indicates that a returned access token would grant access to the given claims at the
UserInfo Endpoint, while the id_token
target indicates that the claims would be returned in an
ID Token as described in Section 3.4.¶
[[ Editor's note: in order to use the "claims" parameter as defined in OIDC, we have to violate the principle of orthogonality in Section 2.9. An alternative approach would be to split up the portions of the claims request, so that "id_token" claims would go into the "subject" field and "userinfo" claims would go into the "resources" request, but this violates the original field definition from OIDC and gets into the territory of defining an identity schema request. This approach would also invalidate extensions to the "claims" standard as each "target" would need to have its own separate mapping to some part of the GNAP protocol. ]]¶
[[ Editor's note: I'm not a fan of GNAP defining how OIDC would work at all and would rather that work be done by the OIDF in an extension. However, I think it is important for discussion to see this kind of thing in context with the rest of the protocol, for now. In the future, I would anticipate this would be defined by the OIDF as a relatively small but robust identity layer on top of GNAP. ]]¶
The request object MAY be extended by registering new items in a registry TBD (Section 12). Extensions SHOULD be orthogonal to other parameters. Extensions MUST document any aspects where the extension item affects or influences the values or behavior of other request and response objects.¶
[[ Editor's note: we should have more guidance and examples on what possible top-level extensions would look like. ]]¶
In response to a RC's request, the AS responds with a JSON object as the HTTP entity body. Each possible field is detailed in the sections below¶
In this example, the AS is returning an interaction URL (Section 3.3.1), a callback nonce (Section 3.3.3), and a continuation handle (Section 3.1).¶
{ "interact": { "redirect": "https://server.example.com/interact/4CF492MLVMSW9MKMXKHQ", "callback": "MBDOFXG4Y5CVJCX821LH" }, "continue": { "access_token": { "value": "80UPRY5NM33OMUKMKSKU", "key": true }, "uri": "https://server.example.com/tx" } }¶
In this example, the AS is returning a bearer access token (Section 3.2.1) with a management URL and a subject identifier (Section 3.4) in the form of an email address.¶
{ "access_token": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": false, "manage": "https://server.example.com/token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L" }, "subject": { "sub_ids": [ { "subject_type": "email", "email": "user@example.com", } ] } }¶
If the AS determines that the request can be continued with additional requests, it responds with the "continue" field. This field contains a JSON object with the following properties.¶
bearer
token.
This access token MUST NOT be usable at resources outside of the AS.
[[ Editor's note: Is this a restriction we want to enforce? ]]
If the AS includes an access token, the RC MUST present the access
token in all requests to the continuation URI as
described in Section 7.¶
{ "continue": { "access_token": { "value": "80UPRY5NM33OMUKMKSKU", "key": true }, "uri": "https://server.example.com/continue", "wait": 60 } }¶
The RC can use the values of this field to continue the
request as described in Section 5. Note that the
RC MUST sign all continuation requests with its key as described
in Section 8. If the AS includes an access_token
, the RC
MUST present the access token in its continuation request.¶
This field SHOULD be returned when interaction is expected, to allow the RC to follow up after interaction has been concluded.¶
[[ Editor's note: The AS can use the optional "access_token" as a credential for the client to manage the grant request itself over time. This is in parallel with access token management as well as RS access in general. If the AS uses the access token, the continuation URL can be static, and potentially even the same as the initial request URL. If the AS does not use an access token here, it needs to use unique URLs in its response and bind the client's key to requests to those URLs - or potentially only allow one request per client at a time. The optionality adds a layer of complexity, but the client behavior is deterministic in all possible cases and it re-uses existing functions and structures instead of inventing something special just to talk to the AS. The optional access token represents a design compromise, but the working group can decide to either require the access token on all requests or to remove the access token functionality and require the security of the continuation requests be based on unique URLs. ]]¶
If the AS has successfully granted one or more access tokens to the RC,
the AS responds with either the access_token
or the multiple_access_token
field. The AS MUST NOT respond with both
the access_token
and multiple_access_token
fields.¶
[[ Editor's note: I really don't like the dichotomy between "access_token" and "multiple_access_tokens" and their being mutually exclusive, and I think we should design away from this pattern toward something less error-prone. ]]¶
If the RC has requested a single access token and the AS has granted that access token, the AS responds with the "access_token" field. The value of this field is an object with the following properties.¶
true
is used,
the token is bound to the key used by the RC (Section 2.3.2) in its request
for access. If the boolean value false
is used,
the token is a bearer token with no key bound to it.
Otherwise, the key MUST be an object or string in a format
described in Section 2.3.2, describing a public key to which the
RC can use the associated private key. The RC MUST be able to
dereference or process the key information in order to be able
to sign the request.¶
The following non-normative example shows a single bearer token with a management URL that has access to three described resources.¶
"access_token": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": false, "manage": "https://server.example.com/token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L", "resources": [ { "type": "photo-api", "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }, "read", "dolphin-metadata" ] }¶
The following non-normative example shows a single access token bound to the RC's key, which was presented using the detached JWS (Section 8.1) binding method.¶
"access_token": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": true, "resources": [ "finance", "medical" ] }¶
If the RC requested multiple access tokens (Section 2.1.3), the AS MUST NOT respond with a
single access token structure unless the RC sends the split_token
flag as described in Section 2.1.4.¶
[[ Editor's note: There has been interest in describing a way for the AS to tell the client both how and where to use the token. This kind of directed access token could allow for some interesting deployment patterns where the client doesn't know much]]¶
If the RC has requested multiple access tokens and the AS has granted at least one of them, the AS responds with the "multiple_access_tokens" field. The value of this field is a JSON object, and the property names correspond to the token identifiers chosen by the RC in the multiple access token request (Section 2.1.3). The values of the properties of this object are access tokens as described in Section 3.2.1.¶
In this non-normative example, two bearer tokens are issued under the
names token1
and token2
, and only the first token has a management
URL associated with it.¶
"multiple_access_tokens": { "token1": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": false, "manage": "https://server.example.com/token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L" }, "token2": { "value": "UFGLO2FDAFG7VGZZPJ3IZEMN21EVU71FHCARP4J1", "key": false } }¶
Each access token corresponds to the named resources arrays in the RC's request (Section 2.1.3).¶
The multiple access token response MUST be used when multiple access tokens are requested, even if only one access token is issued as a result of the request. The AS MAY refuse to issue one or more of the requested access tokens, for any reason. In such cases the refused token is omitted from the response and all of the other issued access tokens are included in the response the requested names appropriate names.¶
If the RC requested a single access token (Section 2.1.1), the AS MUST NOT respond with the multiple
access token structure unless the RC sends the split_token
flag as described in Section 2.1.4.¶
Each access token MAY have different proofing mechanisms. If management is allowed, each access token SHOULD have different management URIs.¶
[[ Editor's note: Do we need to specify that the management URIs are different if we require the token to be presented? ]]¶
If the RC has indicated a capability to interact with the RO in its request (Section 2.5),
and the AS has determined that interaction is both
supported and necessary, the AS responds to the RC with any of the
following values in the interact
field of the response. There is
no preference order for interaction modes in the response,
and it is up to the RC to determine which ones to use. All supported
interaction methods are included in the same interact
object.¶
Additional interaction mode responses can be defined in a registry TBD (Section 12).¶
The AS MUST NOT respond with any interaction mode that the RC did not indicate in its request. The AS MUST NOT respond with any interaction mode that the AS does not support. Since interaction responses include secret or unique information, the AS SHOULD respond to each interaction mode only once in an ongoing request, particularly if the RC modifies its request (Section 5.3).¶
If the RC indicates that it can redirect to an arbitrary URL (Section 2.5.1) and the AS supports this mode for the RC's request, the AS responds with the "redirect" field, which is a string containing the URL to direct the RQ to. This URL MUST be unique for the request and MUST NOT contain any security-sensitive information.¶
"interact": { "redirect": "https://interact.example.com/4CF492MLVMSW9MKMXKHQ" }¶
The interaction URL returned represents a function of the AS but MAY be completely distinct from the URL the RC uses to request access (Section 2), allowing an AS to separate its user-interactive functionality from its back-end security functionality.¶
[[ Editor's note: This is one aspect where the AS might actually be two separate roles. Namely, a delegation server (back end) and interaction server (user-facing).]]¶
The RC sends the RQ to the URL to interact with the AS. The RC MUST NOT alter the URL in any way. The means for the RC to send the RQ to this URL is out of scope of this specification, but common methods include an HTTP redirect, launching the system browser, displaying a scannable code, or printing out the URL in an interactive console.¶
If the RC indicates that it can launch an application URL (Section 2.5.2) and the AS supports this mode for the RC's request, the AS responds with the "app" field, which is a string containing the URL to direct the RQ to. This URL MUST be unique for the request and MUST NOT contain any security-sensitive information.¶
"interact": { "app": "https://app.example.com/launch?tx=4CF492MLV" }¶
The RC launches the URL as appropriate on its platform, and the means for the RC to launch this URL is out of scope of this specification. The RC MUST NOT alter the URL in any way. The RC MAY attempt to detect if an installed application will service the URL being sent before attempting to launch the application URL.¶
[[ Editor's note: This will probably need to be expanded to an object to account for other parameters needed in app2app use cases, like addresses for distributed storage systems, server keys, and the like. Details TBD as people build this out. ]]¶
If the RC indicates that it can receive a post-interaction callback on a URL (Section 2.5.3) and the AS supports this mode for the RC's request, the AS responds with a "callback" field containing a nonce that the RC will use in validating the callback as defined in Section 4.4.1.¶
"interact": { "callback": "MBDOFXG4Y5CVJCX821LH" }¶
[[ Editor's note: This is fairly parallel to the request but it kinda hides the fact that this is a nonce from the AS, not the client. ]]¶
When the RO completes interaction at the AS, the AS MUST call the RC's callback URL using the method indicated in the callback request (Section 2.5.3) as described in Section 4.4.1.¶
If the AS returns a "callback" nonce, the RC MUST NOT continue a grant request before it receives the associated interaction reference on the callback URI.¶
If the RC indicates that it can display a short user-typeable code (Section 2.5.4) and the AS supports this mode for the RC's request, the AS responds with a "user_code" field. This field is an object that contains the following members.¶
"interact": { "user_code": { "code": "A1BC-3DFF", "url": "https://srv.ex/device" } }¶
The RC MUST communicate the "code" to the RQ in some
fashion, such as displaying it on a screen or reading it out
audibly. The code
is a one-time-use credential that the AS uses to identify
the pending request from the RC. When the RO enters this code (Section 4.2) into the
AS, the AS MUST determine the pending request that it was associated
with. If the AS does not recognize the entered code, the AS MUST
display an error to the user. If the AS detects too many unrecognized codes
entered, it SHOULD display an error to the user.¶
The RC SHOULD also communicate the URL if possible to facilitate user interaction, but since the URL should be stable, the RC should be able to safely decide to not display this value. As this interaction mode is designed to facilitate interaction via a secondary device, it is not expected that the RC redirect the RQ to the URL given here at runtime. Consequently, the URL needs to be stable enough that a RC could be statically configured with it, perhaps referring the RQ to the URL via documentation instead of through an interactive means. If the RC is capable of communicating an arbitrary URL to the RQ, such as through a scannable code, the RC can use the "redirect" (Section 2.5.1) mode for this purpose instead of or in addition to the user code mode.¶
The interaction URL returned represents a function of the AS but MAY be completely distinct from the URL the RC uses to request access (Section 2), allowing an AS to separate its user-interactive functionality from its back-end security functionality.¶
[[ Editor's note: This is one aspect where the AS might actually be two separate roles. Namely, a delegation server (back end) and interaction server (user-facing).]]¶
Extensions to this specification can define new interaction mode responses in a registry TBD (Section 12). Extensions MUST document the corresponding interaction request.¶
If information about the RO is requested and the AS grants the RC access to that data, the AS returns the approved information in the "subject" response field. This field is an object with the following OPTIONAL properties.¶
"subject": { "sub_ids": [ { "subject_type": "email", "email": "user@example.com", } ], "assertions": { "id_token": "eyj..." } }¶
The AS MUST return the subject
field only in cases where the AS is sure that
the RO and the RQ are the same party. This can be accomplished through some forms of
interaction with the RO (Section 4).¶
Subject identifiers returned by the AS SHOULD uniquely identify the RO at the AS. Some forms of subject identifier are opaque to the RC (such as the subject of an issuer and subject pair), while others forms (such as email address and phone number) are intended to allow the RC to correlate the identifier with other account information at the RC. The RC MUST NOT request or use any returned subject identifiers for communication purposes (see Section 2.2). That is, a subject identifier returned in the format of an email address or a phone number only identifies the RO to the AS and does not indicate that the AS has validated that the represented email address or phone number in the identifier is suitable for communication with the current user. To get such information, the RC MUST use an identity protocol to request and receive additional identity claims. While Section 2.8 specifies one such method, other identity protocols could also be used on top of GNAP to convey this information and the details of an identity protocol and associated schema are outside the scope of this specification.¶
[[ Editor's note: subject identifiers here are naturally scoped to the AS; even though using an external identifier like an email address or phone number implies a global namespace in use, the association of that identifier to the current user is still under the view of the AS. Would changing the name to "as_sub_ids" or "local_sub_ids" help convey that point? Would it also be desirable to have an identifier that's globally unique by design? The "iss_sub" type almost gets us there by explicitly calling out the issuer URL, but tuples are hard to deal with in practice and so tend to get ignored in practice in the OIDC space. ]]¶
[[ Editor's note: This will need substantial privacy considerations, as this is releasing information about the current user that could be tied to other information at the RC or elsewhere. To facilitate this, should we have another form of identifier that's a globally unique identifier of some form? DIDs could facilitate that kind of namespace. ]]¶
Extensions to this specification MAY define additional response properties in a registry TBD (Section 12).¶
Many parts of the RC's request can be passed as either a value or a reference. The use of a reference in place of a value allows for a client to optimize requests to the AS.¶
Some references, such as for the RC instance's identity (Section 2.3.1) or the requested resources (Section 2.1.2), can be managed statically through an admin console or developer portal provided by the AS or RS. The developer of the RC can include these values in their code for a more efficient and compact request.¶
If desired, the AS MAY also generate and return some of these references dynamically to the RC in its response to facilitate multiple interactions with the same software. The RC SHOULD use these references in future requests in lieu of sending the associated data value. These handles are intended to be used on future requests.¶
Dynamically generated handles are string values that MUST be protected by the RC as secrets. Handle values MUST be unguessable and MUST NOT contain any sensitive information. Handle values are opaque to the RC.¶
[[ Editor's note: these constructs used to be objects to allow for expansion to future fields, like a management URI or different presentation types or expiration, but those weren't used in practice. Is that desirable anymore or is collapsing them like this the right direction? ]]¶
All dynamically generated handles are returned as fields in the root JSON object of the response. This specification defines the following dynamic handle returns, additional handles can be defined in a registry TBD (Section 12).¶
client
object that the RC can use in a future request, as
described in Section 2.3.1.¶
This non-normative example shows two handles along side an issued access token.¶
{ "user_handle": "XUT2MFM1XBIKJKSDU8QM", "instance_id": "7C7C4AZ9KHRS6X63AJAO", "access_token": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": false } }¶
[[ Editor's note: the ability to dynamically return reference handles allows for an inline version of dynamic registration without needing to go through a discrete registration step, for clients where that makes sense. Currently this is entirely up to the AS to decide when to issue these, but maybe the client should signal that it can receive these handles as part of the request? The new "token flags" construct in Section 2.1.4 almost gets at that, but for a different part of the request structure. Since the client is the component that will know if it's in a position to make use of such reference handles in the future (like a mobile app) or if it's just going to evaporate at the end of a session (like an SPA). Ultimately we need to deal with a range of dynamism, not just the "pre-registered" vs. "non-registered" use cases that OAuth forces us in to. ]]¶
[[ Editor's note: The client-bound "instance_id" could serve as the hook we would need for RFC7592 style dynamic client management, including additional components like key rotation. If the AS returns an object instead of a string here, that could include everything that the client would need in order to make REST-style management calls, similar to token management.¶
{ "client": { "instance_id": "7C7C4AZ9KHRS6X63AJAO", "manage": "https://example.server.com/client/7C7C4AZ9KHRS6X63AJAO", "access_token": { "value": "4TB8N6BW7OZB8CDFONP219RP1LT0OS9M2PMHKUR6", "key": true } } }¶
The client would sign all requests with its key and use the presented access token. A "POST" or "PATCH" request would update client information, including having a method for key rotation using nested signatures. A "DELETE" request would un-register the client, etc. ]]¶
If the AS determines that the request cannot be issued for any reason, it responds to the RC with an error message.¶
{ "error": "user_denied" }¶
The error code is one of the following, with additional values available in a registry TBD (Section 12):¶
[[ Editor's note: I think we will need a more robust error mechanism, and we need to be more clear about what error states are allowed in what circumstances. Additionally, is the "error" parameter exclusive with others in the return? ]]¶
Extensions to this specification MAY define additional fields for the grant response in a registry TBD (Section 12).¶
[[ Editor's note: what guidance should we give to designers on this? ]]¶
If the RC indicates that it is capable of driving interaction with the RO in its request (Section 2.5), and the AS determines that interaction is required and responds to one or more of the RC's interaction modes, the RC SHOULD initiate one of the returned interaction modes in the response (Section 3.3).¶
When the RO is interacting with the AS, the AS MAY perform whatever actions it sees fit, including but not limited to:¶
[[ Editor's note: there are some privacy and security considerations here but for the most part we don't want to be overly prescriptive about the UX, I think. ]]¶
When the RO is directed to the AS through the "redirect" (Section 3.3.1) mode, the AS can interact with the RO through their web browser to authenticate the user as an RO and gather their consent. Note that since the RC does not add any parameters to the URL, the AS MUST determine the grant request being referenced from the URL value itself. If the URL cannot be associated with a currently active request, the AS MUST display an error to the RO and MUST NOT attempt to redirect the RO back to any RC even if a callback is supplied (Section 2.5.3).¶
The interaction URL MUST be reachable from the RO's browser, though note that the RO MAY open the URL on a separate device from the RC itself. The interaction URL MUST be accessible from an HTTP GET request, and MUST be protected by HTTPS or equivalent means.¶
With this method, it is common for the RO to be the same party as the RQ, since the RC has to communicate the redirection URI to the RQ.¶
When the RO is directed to the AS through the "user_code" (Section 3.3.4) mode, the AS can interact with the RO through their web browser to collect the user code, authenticate the user as an RO, and gather their consent. Note that since the URL itself is static, the AS MUST determine the grant request being referenced from the user code value itself. If the user code cannot be associated with a currently active request, the AS MUST display an error to the RO and MUST NOT attempt to redirect the RO back to any RC even if a callback is supplied (Section 2.5.3).¶
The user code URL MUST be reachable from the RO's browser, though note that the RO MAY open the URL on a separate device from the RC itself. The user code URL MUST be accessible from an HTTP GET request, and MUST be protected by HTTPS or equivalent means.¶
While it is common for the RO to be the same party as the RQ, since the RC has to communicate the user code to someone, there are cases where the RQ and RO are separate parties and the authorization happens asynchronously.¶
When the RC successfully launches an application through the "app" mode (Section 3.3.2), the AS interacts with the RO through that application to authenticate the user as the RO and gather their consent. The details of this interaction are out of scope for this specification.¶
[[ Editor's note: Should we have anything to say about an app sending information to a back-end to get details on the pending request? ]]¶
Upon completing an interaction with the RO, if a "callback" (Section 3.3.3) mode is available with the current request, the AS MUST follow the appropriate method at the end of interaction to allow the RC to continue. If this mode is not available, the AS SHOULD instruct the RO to return to their RC software upon completion. Note that these steps still take place in most error cases, such as when the RO has denied access. This pattern allows the RC to potentially recover from the error state without restarting the request from scratch by modifying its request or providing additional information directly to the AS.¶
[[ Editor's note: there might be some other kind of push-based notification or callback that the client can use, or an out-of-band non-HTTP protocol. The AS would know about this if supported and used, but the guidance here should be written in such a way as to not be too restrictive in the next steps that it can take. Still, it's important that the AS not expect or even allow clients to poll if the client has stated it can take a callback of some form, otherwise that sets up a potential session fixation attack vector that the client is trying to and able to avoid. There has also been a call for post-interaction that doesn't tie into the security of the protocol, like redirecting to a static webpage hosted by the client's company. Would this fit here? ]]¶
The AS MUST create an interaction reference and associate that reference with the current interaction and the underlying pending request. This value MUST be sufficiently random so as not to be guessable by an attacker. The interaction reference MUST be one-time-use.¶
The AS MUST calculate a hash value based on the RC and AS nonces and the interaction reference, as described in Section 4.4.3. The RC will use this value to validate the return call from the AS.¶
The AS then MUST send the hash and interaction reference based on the interaction finalization mode as described in the following sections.¶
When using the "callback" interaction mode (Section 3.3.3) with the redirect
method,
the AS signals to the RC that interaction is
complete and the request can be continued by directing the RO (in
their browser) back to the RC's callback URL sent in the callback request (Section 2.5.3.1).¶
The AS secures this callback by adding the hash and interaction reference as query parameters to the RC's callback URL.¶
The means of directing the RO to this URL are outside the scope of this specification, but common options include redirecting the RO from a web page and launching the system browser with the target URL.¶
https://client.example.net/return/123455 ?hash=p28jsq0Y2KK3WS__a42tavNC64ldGTBroywsWxT4md_jZQ1R2HZT8BOWYHcLmObM7XHPAdJzTZMtKBsaraJ64A &interact_ref=4IFWWIKYBC2PQ6U56NL1¶
When receiving the request, the RC MUST parse the query parameters to calculate and validate the hash value as described in Section 4.4.3. If the hash validates, the RC sends a continuation request to the AS as described in Section 5.1 using the interaction reference value received here.¶
When using the
"callback" interaction mode (Section 3.3.3) with the push
method,
the AS signals to the RC that interaction is
complete and the request can be continued by sending an HTTP POST
request to the RC's callback URL sent in the callback request (Section 2.5.3.2).¶
The entity message body is a JSON object consisting of the following two fields:¶
POST /push/554321 HTTP/1.1 Host: client.example.net Content-Type: application/json { "hash": "p28jsq0Y2KK3WS__a42tavNC64ldGTBroywsWxT4md_jZQ1R2HZT8BOWYHcLmObM7XHPAdJzTZMtKBsaraJ64A", "interact_ref": "4IFWWIKYBC2PQ6U56NL1" }¶
When receiving the request, the RC MUST parse the JSON object and validate the hash value as described in Section 4.4.3. If the hash validates, the RC sends a continuation request to the AS as described in Section 5.1 using the interaction reference value received here.¶
The "hash" parameter in the request to the RC's callback URL ties the front channel response to an ongoing request by using values known only to the parties involved. This security mechanism allows the RC to protect itself against several kinds of session fixation and injection attacks. The AS MUST always provide this hash, and the RC MUST validate the hash when received.¶
[[ Editor's note: If the client uses a unique callback URL per request, that prevents some of the same attacks, but without the same cryptographic binding between the interaction and delegation channels. A unique URI would allow the client to differentiate inputs, but it would not prevent an attacker from injecting an unrelated interaction reference into this channel. ]]¶
To calculate the "hash" value, the party doing the calculation first takes the "nonce" value sent by the RC in the interaction section of the initial request (Section 2.5.3), the AS's nonce value from the callback response (Section 3.3.3), and the "interact_ref" sent to the RC's callback URL. These three values are concatenated to each other in this order using a single newline character as a separator between the fields. There is no padding or whitespace before or after any of the lines, and no trailing newline character.¶
VJLO6A4CAYLBXHTR0KRO MBDOFXG4Y5CVJCX821LH 4IFWWIKYBC2PQ6U56NL1¶
The party then hashes this string with the appropriate algorithm based on the "hash_method" parameter of the "callback". If the "hash_method" value is not present in the RC's request, the algorithm defaults to "sha3".¶
[[ Editor's note: these hash algorithms should be pluggable, and ideally we shouldn't redefine yet another crypto registry for this purpose, but I'm not convinced an appropriate one already exists. Furthermore, we should be following best practices here whether it's a plain hash, a keyed MAC, an HMAC, or some other form of cryptographic function. I'm not sure what the defaults and options ought to be, but SHA512 and SHA3 were picked based on what was available to early developers. ]]¶
The "sha3" hash method consists of hashing the input string with the 512-bit SHA3 algorithm. The byte array is then encoded using URL Safe Base64 with no padding. The resulting string is the hash value.¶
p28jsq0Y2KK3WS__a42tavNC64ldGTBroywsWxT4md_jZQ1R2HZT8BOWYHcLmObM7XHPAdJzTZMtKBsaraJ64A¶
The "sha2" hash method consists of hashing the input string with the 512-bit SHA2 algorithm. The byte array is then encoded using URL Safe Base64 with no padding. The resulting string is the hash value.¶
62SbcD3Xs7L40rjgALA-ymQujoh2LB2hPJyX9vlcr1H6ecChZ8BNKkG_HrOKP_Bpj84rh4mC9aE9x7HPBFcIHw¶
While it is possible for the AS to return a Section 3 with all the RC's requested information (including access tokens (Section 3.2) and direct user information (Section 3.4)), it's more common that the AS and the RC will need to communicate several times over the lifetime of an access grant. This is often part of facilitating interaction (Section 4), but it could also be used to allow the AS and RC to continue negotiating the parameters of the original grant request (Section 2).¶
To enable this ongoing negotiation, the AS returns a continue
field
in the response (Section 3.1) that contains information the RC needs to
continue this process with another request, including a URI to access
as well as an optional access token to use during the continued requests.¶
When the RC makes any calls to the continuation URL, the RC MUST present proof of the most recent key associated with this ongoing request by signing the request as described in Section 8. The key in use will be either the key from the initial request (Section 2.3.2) or its most recent rotation. [[ Editor's note: we need to have a secure way to rotate the key used for the continuation here. In most cases this will be a rotation for the client instance, since a client without an instance record would likely just present a new key for a new request. In that case it could go with the client management, above - but it doesn't necessarily have to be. ]]¶
For example, here the RC makes a POST request and signs with detached JWS:¶
POST /continue/80UPRY5NM33OMUKMKSKU HTTP/1.1 Host: server.example.com Detached-JWS: ejy0...¶
If the AS includes an access_token
in the continue
response in Section 3.1, the RC MUST include the access token the
request as described in Section 7. Note that the access token
is always bound to the RC's presented key (or its most recent rotation).¶
For example, here the RC makes a POST request with the interaction reference, includes the access token, and signs with detached JWS:¶
POST /continue HTTP/1.1 Host: server.example.com Content-type: application/json Authorization: GNAP 80UPRY5NM33OMUKMKSKU Detached-JWS: ejy0... { "interact_ref": "4IFWWIKYBC2PQ6U56NL1" }¶
The AS MUST be able to tell from the RC's request which specific ongoing request is being accessed. Common methods for doing so include using a unique, unguessable URL for each continuation response, associating the request with the provided access token, or allowing only a single ongoing grant request for a given RC instance at a time. If the AS cannot determine a single active grant request to map the continuation request to, the AS MUST return an error.¶
The ability to continue an already-started request allows the RC to perform several important functions, including presenting additional information from interaction, modifying the initial request, and getting the current state of the request.¶
If a "wait" parameter was included in the continuation response (Section 3.1), the RC MUST NOT call the continuation URI prior to waiting the number of seconds indicated. If no "wait" period is indicated, the RC SHOULD wait at least 5 seconds [[ Editor's note: what's a reasonable amount of time so as not to DOS the server?? ]]. If the RC does not respect the given wait period, the AS MUST return an error.¶
The response from the AS is a JSON object and MAY contain any of the fields described in Section 3, as described in more detail in the sections below.¶
If the AS determines that the RC can
make a further continuation request, the AS MUST include a new
"continue" response (Section 3.1).
If the continuation was previously bound to an access token, the
new continue
response MUST include a bound access token as well, and
this token SHOULD be a new access token. [[ Editor's note: this used to be a MUST,
but is it safe to back off that requirement? ]]
If the AS does not return a new continue
response, the RC
MUST NOT make an additional continuation request. If a RC does so,
the AS MUST return an error.¶
For continuation functions that require the RC to send a message body, the body MUST be a JSON object.¶
When the AS responds to the RC's callback
parameter as in Section 4.4.1, this
response includes an interaction reference. The RC MUST include that value as the field
interact_ref
in a POST request to the continuation URI.¶
POST /continue/80UPRY5NM33OMUKMKSKU HTTP/1.1 Host: server.example.com Content-type: application/json Detached-JWS: ejy0... { "interact_ref": "4IFWWIKYBC2PQ6U56NL1" }¶
Since the interaction reference is a one-time-use value as described in Section 4.4.1, if the RC needs to make additional continuation calls after this request, the RC MUST NOT include the interaction reference. If the AS detects an RC submitting the same interaction reference multiple times, the AS MUST return an error and SHOULD invalidate the ongoing request.¶
The Section 3 MAY contain any newly-created access tokens (Section 3.2) or newly-released subject claims (Section 3.4). The response MAY contain a new "continue" response (Section 3.1) as described above. The response SHOULD NOT contain any interaction responses (Section 3.3). [[ Editor's note: This last one might be overly restrictive, since some kinds of interaction could require multiple round trips. We need more examples and experience beyond redirect-based interaction here. ]]¶
For example, if the request is successful in causing the AS to issue access tokens and release subject claims, the response could look like this:¶
{ "access_token": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": false, "manage": "https://server.example.com/token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L" }, "subject": { "sub_ids": [ { "subject_type": "email", "email": "user@example.com", } ] } }¶
With this example, the RC can not make an additional continuation request because
a continue
field is not included.¶
[[ Editor's note: other interaction methods, such as a challenge-response cryptographic protocol, would use a similar construct as here, but have different rules. Would it be reasonable to allow them to be combined? Could this be combined further with the "update" method in Section 5.3? ]]¶
When the RC does not include a callback
parameter, the RC will often need to
poll the AS until the RO has authorized the request. To do so, the RC makes a POST
request to the continuation URI as in Section 5.1, but does not
include a message body.¶
POST /continue HTTP/1.1 Host: server.example.com Content-type: application/json Authorization: GNAP 80UPRY5NM33OMUKMKSKU Detached-JWS: ejy0...¶
The Section 3 MAY contain any newly-created access tokens (Section 3.2) or
newly-released subject claims (Section 3.4). The response MAY contain
a new "continue" response (Section 3.1) as described above. If a continue
field is included, it SHOULD include a wait
field to facilitate a reasonable polling rate by
the RC. The response SHOULD NOT contain interaction responses (Section 3.3).¶
For example, if the request has not yet been authorized by the RO, the AS could respond by telling the RC to make another continuation request in the future. In this example, a new, unique access token has been issued for the call, which the RC will use in its next continuation request.¶
{ "continue": { "access_token": { "value": "33OMUKMKSKU80UPRY5NM", "key": true }, "uri": "https://server.example.com/continue", "wait": 30 } }¶
[[ Editor's note: Do we want to be more precise about what's expected inside the "continue" object? I think that at least the URI is required, access token required IF used, etc. This is even if they haven't changed since last time, and the client will use whatever value comes back. ]]¶
[[ Editor's note: extensions to this might need to communicate to the client what the current state of the user interaction is. This has been done in similar proprietary protocols, but the details of that information tend to be highly application specific. Like "user hasn't logged in yet", "user has logged in but is still sitting at the page", or "user seems to have wandered off". We might be able to provide a decent framework for hanging this kind of stuff on. ]]¶
If the request is successful in causing the AS to issue access tokens and release subject claims, the response could look like this example:¶
{ "access_token": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": false, "manage": "https://server.example.com/token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L" }, "subject": { "sub_ids": [ { "subject_type": "email", "email": "user@example.com", } ] } }¶
The RC might need to modify an ongoing request, whether or not tokens have already been issued or claims have already been released. In such cases, the RC makes an HTTP PATCH request to the continuation URI and includes any fields it needs to modify. Fields that aren't included in the request are considered unchanged from the original request.¶
The RC MAY include the resources
and subject
fields as described in Section 2.1
and Section 2.2. Inclusion of these fields override any values in the initial request,
which MAY trigger additional requirements and policies by the AS. For example, if the RC is asking for
more access, the AS could require additional interaction with the RO to gather additional consent.
If the RC is asking for more limited access, the AS could determine that sufficient authorization
has been granted to the RC and return the more limited access rights immediately.
[[ Editor's note: We could state something like "resources and subject MUST NOT be the same as in
the initial or previous request" to enforce that this really is a change, but is there value in
calling that out here? Somehow we do probably want to tell the AS to not let a client simply
post the same request here to rotate access tokens now that we've got an explicit function for that, right? ]]¶
The RC MAY include the interact
field as described in Section 2.5. Inclusion of
this field indicates that the RC is capable of driving interaction with the RO, and this field
replaces any values from a previous request. The AS MAY respond to any of the interaction
responses as described in Section 3.3, just like it would to a new request.¶
The RC MAY include the user
field as described in Section 2.4 to present new assertions
or information about the RQ. [[ Editor's note: This would allow the client to do things like
gather the user's identifiers post-request, or gather an assertion from an on-device element
that the AS can verify. It opens up potential avenues for trouble if the user here is
different from the RO that's already showed up at the AS or race conditions if the RQ's identity changes
mid-stream. But that said, this seems important for multi-log-in cases and the like, probably. ]]¶
The RC MUST NOT include the client
section of the request. [[ Editor's note: We do not want
the client to be able to get swapped out from underneath the user, especially post-consent.
However, including this field in a PATCH update request might be the place to
define key rotation for the grant request itself, but we'd need to be very careful of how that works.
And it feels like it might have consequences outside of the request, such as rotating the
key for all ongoing grants for a given client instance, which isn't really desirable here.
We need a lot more discussion and engineering on this before including it. ]]¶
The RC MAY include post-interaction responses such as described in Section 5.1. [[ Editor's note: it seems a little odd to include this in a request but I can't see a reason to not allow it. ]]¶
Modification requests MUST NOT alter previously-issued access tokens. Instead, any access tokens issued from a continuation are considered new, separate access tokens. The AS MAY revoke existing access tokens after a modification has occurred. [[ Editor's note: this might be subject to the "multi_token" flag, but since we're creating a NEW access token and not rotating an existing one, this seems to be a different use case. ]]¶
Modification requests MAY result in previously-issued access tokens being revoked. [[ Editor's note: there is a solid argument to be made for always revoking old access tokens here, but we need more discussion on the boundaries for such a requirement. If they stick around, it does make a "read" request weird because now we've got multiple access tokens sticking around associated with a grant request and no good place to put them. ]]¶
If the modified request can be granted immediately by the AS, the Section 3 MAY contain any newly-created access tokens (Section 3.2) or newly-released subject claims (Section 3.4). The response MAY contain a new "continue" response (Section 3.1) as described above. If interaction can occur, the response SHOULD contain interaction responses (Section 3.3) as well.¶
For example, an RC initially requests a set of resources using references:¶
POST /tx HTTP/1.1 Host: server.example.com Content-type: application/json Detached-JWS: ejy0... { "resources": [ "read", "write" ], "interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.example.net/return/123455", "nonce": "LKLTI25DK82FX4T4QFZC" } }, "client": "987YHGRT56789IOLK" }¶
Access is granted by the RO, and a token is issued by the AS.
In its final response, the AS includes a continue
field:¶
{ "continue": { "access_token": { "value": "80UPRY5NM33OMUKMKSKU", "key": true }, "uri": "https://server.example.com/continue", "wait": 30 }, "access_token": ... }¶
This allows the RC to make an eventual continuation call. The RC realizes that it no longer needs "write" access and therefore modifies its ongoing request, here asking for just "read" access instead of both "read" and "write" as before.¶
PATCH /continue HTTP/1.1 Host: server.example.com Content-type: application/json Authorization: GNAP 80UPRY5NM33OMUKMKSKU Detached-JWS: ejy0... { "resources": [ "read" ] ... }¶
The AS replaces the previous resources
from the first request, allowing the AS to
determine if any previously-granted consent already applies. In this case, the AS would
likely determine that reducing the breadth of the requested access means that new access
tokens can be issued to the RC. The AS would likely revoke previously-issued access tokens
that had the greater access rights associated with them.¶
{ "continue": { "access_token": { "value": "M33OMUK80UPRY5NMKSKU", "key": true }, "uri": "https://server.example.com/continue", "wait": 30 }, "access_token": ... }¶
For another example, the RC initially requests read-only access but later needs to step up its access. The initial request could look like this example.¶
POST /tx HTTP/1.1 Host: server.example.com Content-type: application/json Detached-JWS: ejy0... { "resources": [ "read" ], "interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.example.net/return/123455", "nonce": "LKLTI25DK82FX4T4QFZC" } }, "client": "987YHGRT56789IOLK" }¶
Access is granted by the RO, and a token is issued by the AS.
In its final response, the AS includes a continue
field:¶
{ "continue": { "access_token": { "value": "80UPRY5NM33OMUKMKSKU", "key": true }, "uri": "https://server.example.com/continue", "wait": 30 }, "access_token": ... }¶
This allows the RC to make an eventual continuation call. The RC later realizes that it now needs "write" access in addition to the "read" access. Since this is an expansion of what it asked for previously, the RC also includes a new interaction section in case the AS needs to interact with the RO again to gather additional authorization. Note that the RC's nonce and callback are different from the initial request. Since the original callback was already used in the initial exchange, and the callback is intended for one-time-use, a new one needs to be included in order to use the callback again.¶
[[ Editor's note: the net result of this is that interaction requests are really only meant to be responded to exactly once by the AS. This isn't spelled out explicitly, but could be included in Section 2.5 and/or Section 3.3. ]]¶
PATCH /continue HTTP/1.1 Host: server.example.com Content-type: application/json Authorization: GNAP 80UPRY5NM33OMUKMKSKU Detached-JWS: ejy0... { "resources": [ "read", "write" ], "interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.example.net/return/654321", "nonce": "K82FX4T4LKLTI25DQFZC" } } }¶
From here, the AS can determine that the RC is asking for more than it was previously granted, but since the RC has also provided a mechanism to interact with the RO, the AS can use that to gather the additional consent. The protocol continues as it would with a new request. Since the old access tokens are good for a subset of the rights requested here, the AS might decide to not revoke them. However, any access tokens granted after this update process are new access tokens and do not modify the rights of existing access tokens.¶
If the RC needs to get the current state of an ongoing grant request, it makes an HTTP GET request to the continuation URI. This request MUST NOT alter the grant request in any fashion, including causing the issuance of new access tokens or modification of interaction parameters.¶
The AS MAY include existing access tokens and previously-released subject claims in the response. The AS MUST NOT issue a new access token or release a new subject claim in response to this request.¶
GET /continue HTTP/1.1 Host: server.example.com Content-type: application/json Authorization: GNAP 80UPRY5NM33OMUKMKSKU Detached-JWS: ejy0...¶
The response MAY include any fields described Section 3 that are applicable to this ongoing request, including the most recently issued access tokens, any released subject claims, and any currently active interaction modes. The response MAY contain a new "continue" response (Section 3.1) as described above.¶
[[ Editor's note: I'm a little dubious about the need for this particular function in reality, but including it for completeness sake. There are a lot of questions we need to answer, such as whether it's safe to include access tokens and claims in the response of this kind of "read" at all, and whether it makes sense to include items like interaction nonces in the response. This discussion should be driven by the use cases calling for this "read" functionality. There have been similar functions within proprietary protocols where the client calls an endpoint at the AS to figure out where the user is in the interaction process at the AS, letting the client provide a smarter UI. It doesn't seem like we could do that in depth here since it would be highly application specific, but that might be a good example of how to extend a response and give a client extra information. ]]¶
If the RC wishes to cancel an ongoing grant request, it makes an HTTP DELETE request to the continuation URI.¶
DELETE /continue HTTP/1.1 Host: server.example.com Content-type: application/json Authorization: GNAP 80UPRY5NM33OMUKMKSKU Detached-JWS: ejy0...¶
If the request is successfully cancelled, the AS responds with an HTTP 202. The AS MUST revoke all associated access tokens, if possible.¶
If an access token response includes the "manage" parameter as described in Section 3.2.1, the RC MAY call this URL to manage the access token with any of the actions defined in the following sections. Other actions are undefined by this specification.¶
The access token being managed acts as the access element for its own management API. The RC MUST present proof of an appropriate key along with the access token.¶
If the token is sender-constrained (i.e., not a bearer token), it MUST be sent with the appropriate binding for the access token (Section 7).¶
If the token is a bearer token, the RC MUST present proof of the same key identified in the initial request (Section 2.3.2) as described in Section 8.¶
The AS MUST validate the proof and assure that it is associated with either the token itself or the RC the token was issued to, as appropriate for the token's presentation type.¶
[[ Editor's note: Should we allow for "update" to an access token by the client posting new information from a "request"? It seems this might make things weird since an access token is generally considered an unchanging thing, and the client could always request a new access token if they're allowed to continue the grant request post-issuance as in Section 5.3. There's also a possibility of being able to "read" a token's state with a GET, much like token introspection but using the token's/client's key instead of the RS key. But would a client need to "read" a token state after issuance? Is there a security risk to offering that functionality? Introspection is nearly always relegated to RS calls in practice since the client is focused on using the token at the RS. The client can always read the state of the grant itself, separately. ]]¶
The RC makes an HTTP POST to the token management URI, sending the access token in the appropriate header and signing the request with the appropriate key.¶
POST /token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L HTTP/1.1 Host: server.example.com Authorization: GNAP OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0 Detached-JWS: eyj0....¶
[[ Editor's note: This could alternatively be an HTTP PUT verb, since we are telling the AS that we want to replace the token. However, we are not providing the information we want to replace the token with, and in fact that's up to the AS entirely, not the client. For that reason, I think a POST still makes the most sense. ]]¶
The AS validates that the token presented is associated with the management URL, that the AS issued the token to the given RC, and that the presented key is appropriate to the token.¶
If the access token has expired, the AS SHOULD honor the rotation request to the token management URL since it is likely that the RC is attempting to refresh the expired token. To support this, the AS MAY apply different lifetimes for the use of the token in management vs. its use at an RS. An AS MUST NOT honor a rotation request for an access token that has been revoked, either by the AS or by the RC through the token management URI (Section 6.2).¶
If the token is validated and the key is appropriate for the
request, the AS MUST invalidate the current access token associated
with this URL, if possible, and return a new access token response as
described in Section 3.2.1, unless the multi_token
flag
is specified in the request. [[ Editor's note: We could also use different
verbs to signal whether or not the old token should be kept around
or not, instead of using a token flag to do this. ]] The value of the
access token MUST NOT be the same as the current value of the access
token used to access the management API. The response MAY include an
updated access token management URL as well, and if so, the RC
MUST use this new URL to manage the new access token.¶
[[ Editor's note: the net result is that the client's always going to use the management URL that comes back. But should we let the server omit it from the response if it doesn't change? That seems like an odd optimization that doesn't help the client. ]]¶
{ "access_token": { "value": "FP6A8H6HY37MH13CK76LBZ6Y1UADG6VEUPEER5H2", "key": false, "manage": "https://server.example.com/token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L", "resources": [ { "type": "photo-api", "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }, "read", "dolphin-metadata" ] } }¶
[[ Editor's note: If the client is using its own key as the proof, like with a bearer access token, the AS is going to need to know if the client's key has been rotated. We don't have a mechanism for rotating the token's key or the client's key yet either - so that could occur through this management function as well. ]]¶
If the RC wishes to revoke the access token proactively, such as when a user indicates to the RC that they no longer wish for it to have access or the RC application detects that it is being uninstalled, the RC can use the token management URI to indicate to the AS that the AS should invalidate the access token for all purposes.¶
The RC makes an HTTP DELETE request to the token management URI, presenting the access token and signing the request with the appropriate key.¶
DELETE /token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L HTTP/1.1 Host: server.example.com Authorization: GNAP OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0 Detached-JWS: eyj0....¶
If the key presented is associated with the token (or the RC, in the case of a bearer token), the AS MUST invalidate the access token, if possible, and return an HTTP 204 response code.¶
204 No Content¶
Though the AS MAY revoke an access token at any time for any reason, the token management function is specifically for the RC's use. If the access token has already expired or has been revoked through other means, the AS SHOULD honor the revocation request to the token management URL as valid, since the end result is still the token not being usable.¶
The method the RC uses to send an access token to the RS depends on the value of the "key" and "proof" parameters in the access token response (Section 3.2.1).¶
If the key value is the boolean false
, the access token is a bearer token
sent using the HTTP Header method defined in [RFC6750].¶
Authorization: Bearer OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0¶
The form parameter and query parameter methods of [RFC6750] MUST NOT be used.¶
If the "key" value is the boolean true
, the access token MUST be sent
to the RS using the same key and proofing mechanism that the RC used
in its initial request.¶
If the "key" value is an object, the value of the "proof" field within the key indicates the particular proofing mechanism to use. The access token is sent using the HTTP authorization scheme "GNAP" along with a key proof as described in Section 8 for the key bound to the access token. For example, a "jwsd"-bound access token is sent as follows:¶
Authorization: GNAP OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0 Detached-JWS: eyj0....¶
[[ Editor's note: I don't actually like the idea of using only one header type for differently-bound access tokens. Perhaps instead these values should somehow reflect the key binding types. Maybe there can be multiple fields after the "GNAP" keyword using structured headers? Or a set of derived headers like GNAP-mtls? This might also be better as a separate specification, like it was in OAuth 2. However, access tokens should be able to use any key binding mechanisms here, plus bearer. ]]¶
Any keys presented by the RC to the AS or RS MUST be validated as part of the request in which they are presented. The type of binding used is indicated by the proof parameter of the key section in the initial request Section 2.3.2. Values defined by this specification are as follows:¶
Additional proofing methods are defined by a registry TBD (Section 12).¶
All key binding methods used by this specification MUST cover all relevant portions of the request, including anything that would change the nature of the request, to allow for secure validation of the request by the AS. Relevant aspects include the URI being called, the HTTP method being used, any relevant HTTP headers and values, and the HTTP message body itself. The recipient of the signed message MUST validate all components of the signed message to ensure that nothing has been tampered with or substituted in a way that would change the nature of the request.¶
When used in the GNAP delegation protocol, these key binding mechanisms allow the AS to ensure that the keys presented by the RC in the initial request are in control of the party calling any follow-up or continuation requests. To facilitate this requirement, all keys in the initial request Section 2.3.2 MUST be proved in all continuation requests Section 5 and token management requests Section 6, modulo any rotations on those keys over time that the AS knows about. The AS MUST validate all keys presented by the RC (Section 2.3.2) or referenced in an ongoing request for each call within that request.¶
[[ Editor's note: We are going to need a way for a client to rotate its keys securely, even while an ongoing grant is in effect. ]]¶
When used to bind to an access token, the¶
This method is indicated by jwsd
in the
proof
field. A JWS [RFC7515] signature object is created as follows:¶
The header of the JWS MUST contain the
kid
field of the key bound to this RC for this request. The JWS header
MUST contain an alg
field appropriate for the key identified by kid
and MUST NOT be none
. The b64
field MUST be set to false
and the
crit
field MUST contain at least b64
as specified in [RFC7797]¶
To protect the request, the JWS header MUST contain the following additional fields.¶
access_token
value, where the hash algorithm used is the hash algorithm used in the alg
header parameter of the JWS's JOSE Header. For instance, if the alg
is RS256
, hash the access_token
value with SHA-256, then take the left-most 128 bits and base64url encode them.¶
[[ Editor's note: It's not the usual practice to put additional information into the header of a JWS, but this keeps us from having to normalize the body serialization. Alternatively, we could add all these fields to the body of the request, but then it gets awkward for non-body requests like GET/DELETE. ]]¶
The payload of the JWS object is the serialized body of the request, and the object is signed according to detached JWS [RFC7797].¶
The RC presents the signature in the Detached-JWS HTTP Header field. [[ Editor's Note: this is a custom header field, do we need this? It seems like the best place to put this. ]]¶
POST /tx HTTP/1.1 Host: server.example.com Content-Type: application/json Detached-JWS: eyJiNjQiOmZhbHNlLCJhbGciOiJSUzI1NiIsImtpZCI6Inh5ei0xIn0. .Y287HMtaY0EegEjoTd_04a4GC6qV48GgVbGKOhHdJnDtD0VuUlVjLfwne8AuUY3U7e8 9zUWwXLnAYK_BiS84M8EsrFvmv8yDLWzqveeIpcN5_ysveQnYt9Dqi32w6IOtAywkNUD ZeJEdc3z5s9Ei8qrYFN2fxcu28YS4e8e_cHTK57003WJu-wFn2TJUmAbHuqvUsyTb-nz YOKxuCKlqQItJF7E-cwSb_xULu-3f77BEU_vGbNYo5ZBa2B7UHO-kWNMSgbW2yeNNLbL C18Kv80GF22Y7SbZt0e2TwnR2Aa2zksuUbntQ5c7a1-gxtnXzuIKa34OekrnyqE1hmVW peQ { "resources": [ "dolphin-metadata" ], "interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.foo", "nonce": "VJLO6A4CAYLBXHTR0KRO" } }, "client": { "proof": "jwsd", "key": { "jwk": { "kty": "RSA", "e": "AQAB", "kid": "xyz-1", "alg": "RS256", "n": "kOB5rR4Jv0GMeLaY6_It_r3ORwdf8ci_JtffXyaSx8 xYJCNaOKNJn_Oz0YhdHbXTeWO5AoyspDWJbN5w_7bdWDxgpD-y6jnD1u9YhBOCWObNPF vpkTM8LC7SdXGRKx2k8Me2r_GssYlyRpqvpBlY5-ejCywKRBfctRcnhTTGNztbbDBUyD SWmFMVCHe5mXT4cL0BwrZC6S-uu-LAx06aKwQOPwYOGOslK8WPm1yGdkaA1uF_FpS6LS 63WYPHi_Ap2B7_8Wbw4ttzbMS_doJvuDagW8A1Ip3fXFAHtRAcKw7rdI4_Xln66hJxFe kpdfWdiPQddQ6Y1cK2U3obvUg7w" } } "display": { "name": "My Client Display Name", "uri": "https://example.net/client" }, } }¶
If the request being made does not have a message body, such as an HTTP GET, OPTIONS, or DELETE method, the JWS signature is calculated over an empty payload.¶
When the server (AS or RS) receives the Detached-JWS header, it MUST parse its contents as a detached JWS object. The HTTP Body is used as the payload for purposes of validating the JWS, with no transformations.¶
[[ Editor's note: this is a potentially fragile signature mechanism. It doesn't protect arbitrary headers or other specific aspects of the request, but it's simple to calculate and useful for body-driven requests, like the client to the AS. Additionally it is potentially fragile since a multi-tier system could parse the payload and pass the parsed payload downstream with potential transformations, making downstream signature validation impossible. We might want to remove this in favor of general-purpose HTTP signing, or at least provide guidance on its use. ]]¶
This method is indicated by jws
in the
proof
field. A JWS [RFC7515] signature object is created as follows:¶
The header of the JWS MUST contain the
kid
field of the key bound to this RC for this request. The JWS header
MUST contain an alg
field appropriate for the key identified by kid
and MUST NOT be none
.¶
To protect the request, the JWS header MUST contain the following additional fields.¶
access_token
value, where the hash algorithm used is the hash algorithm used in the alg
header parameter of the JWS's JOSE Header. For instance, if the alg
is RS256
, hash the access_token
value with SHA-256, then take the left-most 128 bits and base64url encode them.¶
[[ Editor's note: It's not the usual practice to put additional information into the header of a JWS, but this keeps us from having to modify the body to use this signature method. Alternatively, we could add all these fields to the body of the request, but then it gets awkward for non-body requests like GET/DELETE. ]]¶
The payload of the JWS object is the JSON serialized body of the request, and the object is signed according to JWS and serialized into compact form [RFC7515].¶
The RC presents the JWS as the body of the request along with a
content type of application/jose
. The AS
MUST extract the payload of the JWS and treat it as the request body
for further processing.¶
POST /tx HTTP/1.1 Host: server.example.com Content-Type: application/jose eyJiNjQiOmZhbHNlLCJhbGciOiJSUzI1NiIsImtpZCI6Inh5ei0xIn0.ewogICAgIm NsaWVudCI6IHsKICAgICAgICAibmFtZSI6ICJNeSBDbGllbnQgRGlzcGxheSBOYW1l IiwKICAgICAgICAidXJpIjogImh0dHBzOi8vZXhhbXBsZS5uZXQvY2xpZW50IgogIC AgfSwKICAgICJyZXNvdXJjZXMiOiBbCiAgICAgICAgImRvbHBoaW4tbWV0YWRhdGEi CiAgICBdLAogICAgImludGVyYWN0IjogewogICAgICAgICJyZWRpcmVjdCI6IHRydW UsCiAgICAgICAgImNhbGxiYWNrIjogewogICAgCQkidXJpIjogImh0dHBzOi8vY2xp ZW50LmZvbyIsCiAgICAJCSJub25jZSI6ICJWSkxPNkE0Q0FZTEJYSFRSMEtSTyIKIC AgIAl9CiAgICB9LAogICAgImtleXMiOiB7CgkJInByb29mIjogImp3c2QiLAogICAg ICAgICJqd2tzIjogewogICAgICAgICAgICAia2V5cyI6IFsKICAgICAgICAgICAgIC AgIHsKICAgICAgICAgICAgICAgICAgICAia3R5IjogIlJTQSIsCiAgICAgICAgICAg ICAgICAgICAgImUiOiAiQVFBQiIsCiAgICAgICAgICAgICAgICAgICAgImtpZCI6IC J4eXotMSIsCiAgICAgICAgICAgICAgICAgICAgImFsZyI6ICJSUzI1NiIsCiAgICAg ICAgICAgICAgICAgICAgIm4iOiAia09CNXJSNEp2MEdNZUxhWTZfSXRfcjNPUndkZj hjaV9KdGZmWHlhU3g4eFlKQ0NOYU9LTkpuX096MFloZEhiWFRlV081QW95c3BEV0pi TjV3XzdiZFdEeGdwRC15NmpuRDF1OVloQk9DV09iTlBGdnBrVE04TEM3U2RYR1JLeD JrOE1lMnJfR3NzWWx5UnBxdnBCbFk1LWVqQ3l3S1JCZmN0UmNuaFRUR056dGJiREJV eURTV21GTVZDSGU1bVhUNGNMMEJ3clpDNlMtdXUtTEF4MDZhS3dRT1B3WU9HT3NsSz hXUG0xeUdka2FBMXVGX0ZwUzZMUzYzV1lQSGlfQXAyQjdfOFdidzR0dHpiTVNfZG9K dnVEYWdXOEExSXAzZlhGQUh0UkFjS3c3cmRJNF9YbG42NmhKeEZla3BkZldkaVBRZG RRNlkxY0syVTNvYnZVZzd3IgogICAgICAgICAgICAgICAgfQogICAgICAgICAgICBd CiAgICAgICAgfQogICAgfQp9.Y287HMtaY0EegEjoTd_04a4GC6qV48GgVbGKOhHdJ nDtD0VuUlVjLfwne8AuUY3U7e89zUWwXLnAYK_BiS84M8EsrFvmv8yDLWzqveeIpcN 5_ysveQnYt9Dqi32w6IOtAywkNUDZeJEdc3z5s9Ei8qrYFN2fxcu28YS4e8e_cHTK5 7003WJu-wFn2TJUmAbHuqvUsyTb-nzYOKxuCKlqQItJF7E-cwSb_xULu-3f77BEU_v GbNYo5ZBa2B7UHO-kWNMSgbW2yeNNLbLC18Kv80GF22Y7SbZt0e2TwnR2Aa2zksuUb ntQ5c7a1-gxtnXzuIKa34OekrnyqE1hmVWpeQ¶
If the request being made does not have a message body, such as
an HTTP GET, OPTIONS, or DELETE method, the JWS signature is
calculated over an empty payload and passed in the Detached-JWS
header as described in Section 8.1.¶
[[ Editor's note: A downside to this method is that it requires the content type to be something other than application/json, and it doesn't work against an RS without additional profiling since it takes over the request body - plus we have to specify different delivery locations for a GET vs. a POST, for example. Additionally it is potentially fragile like a detached JWS since a multi-tier system could parse the payload and pass the parsed payload downstream with potential transformations. We might want to remove this in favor of general-purpose HTTP signing, or at least provide guidance on its use. ]]¶
This method is indicated by mtls
in the
proof
field. The RC presents its client
certificate during TLS negotiation with the server (either AS or RS).
The AS or RS takes the thumbprint of the client certificate presented
during mutual TLS negotiation and compares that thumbprint to the
thumbprint presented by the RC application as described in
[RFC8705] section 3.¶
POST /tx HTTP/1.1 Host: server.example.com Content-Type: application/json SSL_CLIENT_CERT: MIIEHDCCAwSgAwIBAgIBATANBgkqhkiG9w0BAQsFADCBmjE3MDUGA1UEAwwuQmVz cG9rZSBFbmdpbmVlcmluZyBSb290IENlcnRpZmljYXRlIEF1dGhvcml0eTELMAkG A1UECAwCTUExCzAJBgNVBAYTAlVTMRkwFwYJKoZIhvcNAQkBFgpjYUBic3BrLmlv MRwwGgYDVQQKDBNCZXNwb2tlIEVuZ2luZWVyaW5nMQwwCgYDVQQLDANNVEkwHhcN MTkwNDEwMjE0MDI5WhcNMjQwNDA4MjE0MDI5WjB8MRIwEAYDVQQDDAlsb2NhbGhv c3QxCzAJBgNVBAgMAk1BMQswCQYDVQQGEwJVUzEgMB4GCSqGSIb3DQEJARYRdGxz Y2xpZW50QGJzcGsuaW8xHDAaBgNVBAoME0Jlc3Bva2UgRW5naW5lZXJpbmcxDDAK BgNVBAsMA01USTCCASIwDQYJKoZIhvcNAQEBBQADggEPADCCAQoCggEBAMmaXQHb s/wc1RpsQ6Orzf6rN+q2ijaZbQxD8oi+XaaN0P/gnE13JqQduvdq77OmJ4bQLokq sd0BexnI07Njsl8nkDDYpe8rNve5TjyUDCfbwgS7U1CluYenXmNQbaYNDOmCdHww UjV4kKREg6DGAx22Oq7+VHPTeeFgyw4kQgWRSfDENWY3KUXJlb/vKR6lQ+aOJytk vj8kVZQtWupPbvwoJe0na/ISNAOhL74w20DWWoDKoNltXsEtflNljVoi5nqsmZQc jfjt6LO0T7O1OX3Cwu2xWx8KZ3n/2ocuRqKEJHqUGfeDtuQNt6Jz79v/OTr8puLW aD+uyk6NbtGjoQsCAwEAAaOBiTCBhjAJBgNVHRMEAjAAMAsGA1UdDwQEAwIF4DBs BgNVHREEZTBjgglsb2NhbGhvc3SCD3Rsc2NsaWVudC5sb2NhbIcEwKgBBIERdGxz Y2xpZW50QGJzcGsuaW+GF2h0dHA6Ly90bHNjbGllbnQubG9jYWwvhhNzc2g6dGxz Y2xpZW50LmxvY2FsMA0GCSqGSIb3DQEBCwUAA4IBAQCKKv8WlLrT4Z5NazaUrYtl TF+2v0tvZBQ7qzJQjlOqAcvxry/d2zyhiRCRS/v318YCJBEv4Iq2W3I3JMMyAYEe 2573HzT7rH3xQP12yZyRQnetdiVM1Z1KaXwfrPDLs72hUeELtxIcfZ0M085jLboX hufHI6kqm3NCyCCTihe2ck5RmCc5l2KBO/vAHF0ihhFOOOby1v6qbPHQcxAU6rEb 907/p6BW/LV1NCgYB1QtFSfGxowqb9FRIMD2kvMSmO0EMxgwZ6k6spa+jk0IsI3k lwLW9b+Tfn/daUbIDctxeJneq2anQyU2znBgQl6KILDSF4eaOqlBut/KNZHHazJh { "resources": [ "dolphin-metadata" ], "interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.foo", "nonce": "VJLO6A4CAYLBXHTR0KRO" } }, "client": { "display": { "name": "My Client Display Name", "uri": "https://example.net/client" }, "key": { "proof": "mtls", "cert": "MIIEHDCCAwSgAwIBAgIBATANBgkqhkiG9w0BAQsFADCBmjE3 MDUGA1UEAwwuQmVzcG9rZSBFbmdpbmVlcmluZyBSb290IENlcnRpZmljYXRlIEF1d Ghvcml0eTELMAkGA1UECAwCTUExCzAJBgNVBAYTAlVTMRkwFwYJKoZIhvcNAQkBFg pjYUBic3BrLmlvMRwwGgYDVQQKDBNCZXNwb2tlIEVuZ2luZWVyaW5nMQwwCgYDVQQ LDANNVEkwHhcNMTkwNDEwMjE0MDI5WhcNMjQwNDA4MjE0MDI5WjB8MRIwEAYDVQQD DAlsb2NhbGhvc3QxCzAJBgNVBAgMAk1BMQswCQYDVQQGEwJVUzEgMB4GCSqGSIb3D QEJARYRdGxzY2xpZW50QGJzcGsuaW8xHDAaBgNVBAoME0Jlc3Bva2UgRW5naW5lZX JpbmcxDDAKBgNVBAsMA01USTCCASIwDQYJKoZIhvcNAQEBBQADggEPADCCAQoCggE BAMmaXQHbs/wc1RpsQ6Orzf6rN+q2ijaZbQxD8oi+XaaN0P/gnE13JqQduvdq77Om J4bQLokqsd0BexnI07Njsl8nkDDYpe8rNve5TjyUDCfbwgS7U1CluYenXmNQbaYND OmCdHwwUjV4kKREg6DGAx22Oq7+VHPTeeFgyw4kQgWRSfDENWY3KUXJlb/vKR6lQ+ aOJytkvj8kVZQtWupPbvwoJe0na/ISNAOhL74w20DWWoDKoNltXsEtflNljVoi5nq smZQcjfjt6LO0T7O1OX3Cwu2xWx8KZ3n/2ocuRqKEJHqUGfeDtuQNt6Jz79v/OTr8 puLWaD+uyk6NbtGjoQsCAwEAAaOBiTCBhjAJBgNVHRMEAjAAMAsGA1UdDwQEAwIF4 DBsBgNVHREEZTBjgglsb2NhbGhvc3SCD3Rsc2NsaWVudC5sb2NhbIcEwKgBBIERdG xzY2xpZW50QGJzcGsuaW+GF2h0dHA6Ly90bHNjbGllbnQubG9jYWwvhhNzc2g6dGx zY2xpZW50LmxvY2FsMA0GCSqGSIb3DQEBCwUAA4IBAQCKKv8WlLrT4Z5NazaUrYtl TF+2v0tvZBQ7qzJQjlOqAcvxry/d2zyhiRCRS/v318YCJBEv4Iq2W3I3JMMyAYEe2 573HzT7rH3xQP12yZyRQnetdiVM1Z1KaXwfrPDLs72hUeELtxIcfZ0M085jLboXhu fHI6kqm3NCyCCTihe2ck5RmCc5l2KBO/vAHF0ihhFOOOby1v6qbPHQcxAU6rEb907 /p6BW/LV1NCgYB1QtFSfGxowqb9FRIMD2kvMSmO0EMxgwZ6k6spa+jk0IsI3klwLW 9b+Tfn/daUbIDctxeJneq2anQyU2znBgQl6KILDSF4eaOqlBut/KNZHHazJh" } }¶
[[ Editor's note: This method requires no changes to the HTTP message itself, since the security relies on the TLS layer. However, the application level will need to validate that the certificate key used in the request is the one expected for the specific request. ]]¶
This method is indicated by dpop
in the
proof
field. The RC creates a Demonstration of Proof-of-Possession
signature header as described in [I-D.ietf-oauth-dpop]
section 2. In addition to the required fields, the DPoP body MUST also
contain a digest of the request body:¶
POST /tx HTTP/1.1 Host: server.example.com Content-Type: application/json DPoP: eyJ0eXAiOiJkcG9wK2p3dCIsImFsZyI6IlJTMjU2IiwiandrIjp7Imt0eSI6Il JTQSIsImUiOiJBUUFCIiwia2lkIjoieHl6LWNsaWVudCIsImFsZyI6IlJTMjU2Iiwibi I6Inp3Q1RfM2J4LWdsYmJIcmhlWXBZcFJXaVk5SS1uRWFNUnBablJySWpDczZiX2VteV RrQmtEREVqU3lzaTM4T0M3M2hqMS1XZ3hjUGRLTkdaeUlvSDNRWmVuMU1LeXloUXBMSk cxLW9MTkxxbTdwWFh0ZFl6U2RDOU8zLW9peXk4eWtPNFlVeU5aclJSZlBjaWhkUUNiT1 9PQzhRdWdtZzlyZ05ET1NxcHBkYU5lYXMxb3Y5UHhZdnhxcnoxLThIYTdna0QwMFlFQ1 hIYUIwNXVNYVVhZEhxLU9fV0l2WVhpY2c2STVqNlM0NFZOVTY1VkJ3dS1BbHluVHhRZE 1BV1AzYll4VlZ5NnAzLTdlVEpva3ZqWVRGcWdEVkRaOGxVWGJyNXlDVG5SaG5oSmd2Zj NWakRfbWFsTmU4LXRPcUs1T1NEbEhUeTZnRDlOcWRHQ20tUG0zUSJ9fQ.eyJodHRwX21 ldGhvZCI6IlBPU1QiLCJodHRwX3VyaSI6Imh0dHA6XC9cL2hvc3QuZG9ja2VyLmludGV ybmFsOjk4MzRcL2FwaVwvYXNcL3RyYW5zYWN0aW9uIiwiaWF0IjoxNTcyNjQyNjEzLCJ qdGkiOiJIam9IcmpnbTJ5QjR4N2pBNXl5RyJ9.aUhftvfw2NoW3M7durkopReTvONng1 fOzbWjAlKNSLL0qIwDgfG39XUyNvwQ23OBIwe6IuvTQ2UBBPklPAfJhDTKd8KHEAfidN B-LzUOzhDetLg30yLFzIpcEBMLCjb0TEsmXadvxuNkEzFRL-Q-QCg0AXSF1h57eAqZV8 SYF4CQK9OUV6fIWwxLDd3cVTx83MgyCNnvFlG_HDyim1Xx-rxV4ePd1vgDeRubFb6QWj iKEO7vj1APv32dsux67gZYiUpjm0wEZprjlG0a07R984KLeK1XPjXgViEwEdlirUmpVy T9tyEYqGrTfm5uautELgMls9sgSyE929woZ59elg { "resources": [ "dolphin-metadata" ], "interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.foo", "nonce": "VJLO6A4CAYLBXHTR0KRO" } }, "client": { "display": { "name": "My Client Display Name", "uri": "https://example.net/client" }, "proof": "dpop", "key": { "jwk": { "kty": "RSA", "e": "AQAB", "kid": "xyz-1", "alg": "RS256", "n": "kOB5rR4Jv0GMeLaY6_It_r3ORwdf8ci_JtffXyaSx8xYJ CCNaOKNJn_Oz0YhdHbXTeWO5AoyspDWJbN5w_7bdWDxgpD-y6jnD1u9YhBOCWObNPFvpkTM 8LC7SdXGRKx2k8Me2r_GssYlyRpqvpBlY5-ejCywKRBfctRcnhTTGNztbbDBUyDSWmFMVCH e5mXT4cL0BwrZC6S-uu-LAx06aKwQOPwYOGOslK8WPm1yGdkaA1uF_FpS6LS63WYPHi_Ap2 B7_8Wbw4ttzbMS_doJvuDagW8A1Ip3fXFAHtRAcKw7rdI4_Xln66hJxFekpdfWdiPQddQ6Y 1cK2U3obvUg7w" } } } }¶
[[ Editor's note: this method requires duplication of the key in the header and the request body, which is redundant and potentially awkward. The signature also doesn't protect the body of the request. ]]¶
This method is indicated by httpsig
in
the proof
field. The RC creates an HTTP
Signature header as described in [I-D.ietf-httpbis-message-signatures] section 4. The RC MUST
calculate and present the Digest header as defined in [RFC3230] and include
this header in the signature.¶
POST /tx HTTP/1.1 Host: server.example.com Content-Type: application/json Content-Length: 716 Signature: keyId="xyz-client", algorithm="rsa-sha256", headers="(request-target) digest content-length", signature="TkehmgK7GD/z4jGkmcHS67cjVRgm3zVQNlNrrXW32Wv7d u0VNEIVI/dMhe0WlHC93NP3ms91i2WOW5r5B6qow6TNx/82/6W84p5jqF YuYfTkKYZ69GbfqXkYV9gaT++dl5kvZQjVk+KZT1dzpAzv8hdk9nO87Xi rj7qe2mdAGE1LLc3YvXwNxuCQh82sa5rXHqtNT1077fiDvSVYeced0UEm rWwErVgr7sijtbTohC4FJLuJ0nG/KJUcIG/FTchW9rd6dHoBnY43+3Dzj CIthXpdH5u4VX3TBe6GJDO6Mkzc6vB+67OWzPwhYTplUiFFV6UZCsDEeu Sa/Ue1yLEAMg=="]} Digest: SHA=oZz2O3kg5SEFAhmr0xEBbc4jEfo= { "resources": [ "dolphin-metadata" ], "interact": { "redirect": true, "callback": { "method": "push", "uri": "https://client.foo", "nonce": "VJLO6A4CAYLBXHTR0KRO" } }, "client": { "display": { "name": "My Client Display Name", "uri": "https://example.net/client" }, "proof": "httpsig", "key": { "jwk": { "kty": "RSA", "e": "AQAB", "kid": "xyz-1", "alg": "RS256", "n": "kOB5rR4Jv0GMeLaY6_It_r3ORwdf8ci_J tffXyaSx8xYJCCNaOKNJn_Oz0YhdHbXTeWO5AoyspDWJbN5w_7bdWDxgpD- y6jnD1u9YhBOCWObNPFvpkTM8LC7SdXGRKx2k8Me2r_GssYlyRpqvpBlY5- ejCywKRBfctRcnhTTGNztbbDBUyDSWmFMVCHe5mXT4cL0BwrZC6S-uu-LAx 06aKwQOPwYOGOslK8WPm1yGdkaA1uF_FpS6LS63WYPHi_Ap2B7_8Wbw4ttz bMS_doJvuDagW8A1Ip3fXFAHtRAcKw7rdI4_Xln66hJxFekpdfWdiPQddQ6 Y1cK2U3obvUg7w" } } } }¶
When used to present an access token as in Section 7, the Authorization header MUST be included in the signature.¶
This method is indicated by oauthpop
in
the proof
field. The RC creates an HTTP
Authorization PoP header as described in [I-D.ietf-oauth-signed-http-request] section 4, with the
following additional requirements:¶
POST /tx HTTP/1.1 Host: server.example.com Content-Type: application/json PoP: eyJhbGciOiJSUzI1NiIsImp3ayI6eyJrdHkiOiJSU0EiLCJlIjoi QVFBQiIsImtpZCI6Inh5ei1jbGllbnQiLCJhbGciOiJSUzI1NiIsIm4iO iJ6d0NUXzNieC1nbGJiSHJoZVlwWXBSV2lZOUktbkVhTVJwWm5ScklqQ3 M2Yl9lbXlUa0JrRERFalN5c2kzOE9DNzNoajEtV2d4Y1BkS05HWnlJb0g zUVplbjFNS3l5aFFwTEpHMS1vTE5McW03cFhYdGRZelNkQzlPMy1vaXl5 OHlrTzRZVXlOWnJSUmZQY2loZFFDYk9fT0M4UXVnbWc5cmdORE9TcXBwZ GFOZWFzMW92OVB4WXZ4cXJ6MS04SGE3Z2tEMDBZRUNYSGFCMDV1TWFVYW RIcS1PX1dJdllYaWNnNkk1ajZTNDRWTlU2NVZCd3UtQWx5blR4UWRNQVd QM2JZeFZWeTZwMy03ZVRKb2t2allURnFnRFZEWjhsVVhicjV5Q1RuUmhu aEpndmYzVmpEX21hbE5lOC10T3FLNU9TRGxIVHk2Z0Q5TnFkR0NtLVBtM 1EifX0.eyJwIjoiXC9hcGlcL2FzXC90cmFuc2FjdGlvbiIsImIiOiJxa0 lPYkdOeERhZVBTZnc3NnFjamtqSXNFRmxDb3g5bTU5NFM0M0RkU0xBIiw idSI6Imhvc3QuZG9ja2VyLmludGVybmFsIiwiaCI6W1siQWNjZXB0Iiwi Q29udGVudC1UeXBlIiwiQ29udGVudC1MZW5ndGgiXSwiVjQ2OUhFWGx6S k9kQTZmQU5oMmpKdFhTd3pjSGRqMUloOGk5M0h3bEVHYyJdLCJtIjoiUE 9TVCIsInRzIjoxNTcyNjQyNjEwfQ.xyQ47qy8bu4fyK1T3Ru1Sway8wp6 5rfAKnTQQU92AUUU07I2iKoBL2tipBcNCC5zLH5j_WUyjlN15oi_lLHym fPdzihtt8_Jibjfjib5J15UlifakjQ0rHX04tPal9PvcjwnyZHFcKn-So Y3wsARn-gGwxpzbsPhiKQP70d2eG0CYQMA6rTLslT7GgdQheelhVFW29i 27NcvqtkJmiAG6Swrq4uUgCY3zRotROkJ13qo86t2DXklV-eES4-2dCxf cWFkzBAr6oC4Qp7HnY_5UT6IWkRJt3efwYprWcYouOVjtRan3kEtWkaWr G0J4bPVnTI5St9hJYvvh7FE8JirIg { "resources": [ "dolphin-metadata" ], "interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.foo", "nonce": "VJLO6A4CAYLBXHTR0KRO" } }, "client": { "display": { "name": "My Client Display Name", "uri": "https://example.net/client" }, "proof": "oauthpop", "key": { "jwk": { "kty": "RSA", "e": "AQAB", "kid": "xyz-1", "alg": "RS256", "n": "kOB5rR4Jv0GMeLaY6_It_r3ORwdf8ci_J tffXyaSx8xYJCCNaOKNJn_Oz0YhdHbXTeWO5AoyspDWJbN5w_7bdWDxgpD- y6jnD1u9YhBOCWObNPFvpkTM8LC7SdXGRKx2k8Me2r_GssYlyRpqvpBlY5- ejCywKRBfctRcnhTTGNztbbDBUyDSWmFMVCHe5mXT4cL0BwrZC6S-uu-LAx 06aKwQOPwYOGOslK8WPm1yGdkaA1uF_FpS6LS63WYPHi_Ap2B7_8Wbw4ttz bMS_doJvuDagW8A1Ip3fXFAHtRAcKw7rdI4_Xln66hJxFekpdfWdiPQddQ6 Y1cK2U3obvUg7w" } } } }¶
[[ Editor's note: This is a stale draft from the OAuth working group, but it does at least provide some basic functionality for protecting HTTP messages with a signature. This work is likely to be subsumed by the general-purpose HTTP message signature mechanism in Section 8.5. ]]¶
By design, the protocol minimizes the need for any pre-flight discovery. To begin a request, the RC only needs to know the endpoint of the AS and which keys it will use to sign the request. Everything else can be negotiated dynamically in the course of the protocol.¶
However, the AS can have limits on its allowed functionality. If the RC wants to optimize its calls to the AS before making a request, it MAY send an HTTP OPTIONS request to the grant request endpoint to retrieve the server's discovery information. The AS MUST respond with a JSON document containing the following information:¶
proof
field of the
key section (Section 2.3.2) of the request.¶
The information returned from this method is for optimization
purposes only. The AS MAY deny any request, or any portion of a request,
even if it lists a capability as supported. For example, a given RC
can be registered with the mtls
key proofing
mechanism, but the AS also returns other proofing methods, then the AS
will deny a request from that RC using a different proofing
mechanism.¶
In some deployments, a resource server will need to be able to call the AS for a number of functions.¶
[[ Editor's note: This section is for discussion of possible advanced functionality. It seems like it should be a separate document or set of documents, and it's not even close to being well-baked. This also adds additional endpoints to the AS, as this is separate from the token request process, and therefore would require RS-facing discovery or configuration information to make it work. Also-also, it does presume the RS can sign requests in the same way that a client does, but hopefully we can be more consistent with this than RFC7662 was able to do. ]]¶
When the RS receives an access token, it can call the introspection endpoint at the AS to get token information. [[ Editor's note: this isn't super different from the token management URIs, but the RS has no way to get that URI, and it's bound to the RS's keys instead of the RC's or token's keys. ]]¶
+------+ +------+ +------+ | RC |--(1)->| RS | | AS | | | | |--(2)->| | | | | |<-(3)--| | | | | | +------+ | |<-(4)--| | +------+ +------+¶
The RS signs the request with its own key and sends the access token as the body of the request.¶
POST /introspect HTTP/1.1 Host: server.example.com Content-type: application/json Detached-JWS: ejy0... { "access_token": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", }¶
The AS responds with a data structure describing the token's current state and any information the RS would need to validate the token's presentation, such as its intended proofing mechanism and key material.¶
Content-type: application/json { "active": true, "resources": [ "dolphin-metadata", "some other thing" ], "client": { "key": { "proof": "httpsig", "jwk": { "kty": "RSA", "e": "AQAB", "kid": "xyz-1", "alg": "RS256", "n": "kOB5rR4Jv0GMeL...." } } } }¶
Some architectures require an RS to act as an RC and request a derived access token for a secondary RS. This internal token is issued in the context of the incoming access token.¶
+------+ +-------+ +------+ +-------+ | RC |--(1)->| RS1 | | AS | | RS2 | | | | |--(2)->| | | | | | | |<-(3)--| | | | | | | | +------+ | | | | | | | | | | | |-----------(4)------->| | | | | |<----------(5)--------| | | |<-(6)--| | | | +------+ +-------+ +-------+¶
If the RS needs to derive a token from one presented to it, it can request one from the AS by making a token request as described in Section 2 and presenting the existing access token's value in the "existing_access_token" field.¶
The RS MUST identify itself with its own key and sign the request.¶
[[ Editor's note: this is similar to Section 2.7 but based on the access token and not the grant. We might be able to re-use that function: the fact that the keys presented are not the ones used for the access token should indicate that it's a different party and a different kind of request, but there might be some subtle security issues there. ]]¶
POST /tx HTTP/1.1 Host: server.example.com Content-type: application/json Detached-JWS: ejy0... { "resources": [ { "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }, "dolphin-metadata" ], "client": "7C7C4AZ9KHRS6X63AJAO", "existing_access_token": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0" }¶
If the RS needs to, it can post a set of resources as described in Section 2.1.1 to the AS's resource registration endpoint.¶
The RS MUST identify itself with its own key and sign the request.¶
POST /resource HTTP/1.1 Host: server.example.com Content-type: application/json Detached-JWS: ejy0... { "resources": [ { "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }, "dolphin-metadata" ], "client": "7C7C4AZ9KHRS6X63AJAO" }¶
The AS responds with a handle appropriate to represent the resources list that the RS presented.¶
Content-type: application/json { "resource_handle": "FWWIKYBQ6U56NL1" }¶
The RS MAY make this handle available as part of a response (Section 10.4) or as documentation to developers.¶
[[ Editor's note: It's not an exact match here because the "resource_handle" returned now represents a collection of objects instead of a single one. Perhaps we should let this return a list of strings instead? Or use a different syntax than the resource request? Also, this borrows heavily from UMA 2's "distributed authorization" model and, like UMA, might be better suited to an extension than the core protocol. ]]¶
If the RC calls an RS without an access token, or with an invalid access token, the RS MAY respond to the RC with an authentication header indicating that GNAP. The address of the GNAP endpoint MUST be sent in the "as_uri" parameter. The RS MAY additionally return a resource reference that the RC MAY use in its resource request (Section 2.1). This resource reference handle SHOULD be sufficient for at least the action the RC was attempting to take at the RS. The RS MAY use the dynamic resource handle request (Section 10.3) to register a new resource handle, or use a handle that has been pre-configured to represent what the AS is protecting. The content of this handle is opaque to the RS and the RC.¶
WWW-Authenticate: GNAP as_uri=http://server.example/tx,resource=FWWIKYBQ6U56NL1¶
The RC then makes a call to the "as_uri" as described in Section 2, with the value of "resource" as one of the members of a "resources" array Section 2.1.1. The RC MAY request additional resources and other information, and MAY request multiple access tokens.¶
[[ Editor's note: this borrows heavily from UMA 2's "distributed authorization" model and, like UMA, might be better suited to an extension than the core protocol. ]]¶
The author would like to thank the feedback of the following individuals for their reviews, implementations, and contributions: Aaron Parecki, Annabelle Backman, Dick Hardt, Dmitri Zagidulin, Dmitry Barinov, Fabien Imbault, Francis Pouatcha, George Fletcher, Haardik Haardik, Hamid Massaoud, Jacky Yuan, Joseph Heenan, Kathleen Moriarty, Mike Jones, Mike Varley, Nat Sakimura, Takahiko Kawasaki, Takahiro Tsuchiya.¶
In particular, the author would like to thank Aaron Parecki and Mike Jones for insights into how to integrate identity and authentication systems into the core protocol, and to Dick Hardt for the use cases, diagrams, and insights provided in the XAuth proposal that have been incorporated here. The author would like to especially thank Mike Varley and the team at SecureKey for feedback and development of early versions of the XYZ protocol that fed into this standards work.¶
[[ TBD: There are a lot of items in the document that are expandable through the use of value registries. ]]¶
[[ TBD: There are a lot of security considerations to add. ]]¶
All requests have to be over TLS or equivalent as per [BCP195]. Many handles act as shared secrets, though they can be combined with a requirement to provide proof of a key as well.¶
[[ TBD: There are a lot of privacy considerations to add. ]]¶
Handles are passed between parties and therefore should not contain any private data.¶
When user information is passed to the RC, the AS needs to make sure that it has the permission to do so.¶
-14¶
-13¶
-12¶
-11¶
-10¶
-09¶
-08¶
-07¶
-06¶
-05¶
-04¶
-03¶
-02¶
-01¶
-00¶
While different implementations of this protocol will have different realizations of all the components and artifacts enumerated here, the nature of the protocol implies some common structures and elements for certain components. This appendix seeks to enumerate those common elements.¶
TBD: Client has keys, allowed requested resources, identifier(s), allowed requested subjects, allowed¶
TBD: AS has "grant endpoint", interaction endpoints, store of trusted client keys, policies¶
TBD: Token has RO, user, client, resource list, RS list,¶
The protocol defined in this specification provides a number of features that can be combined to solve many different kinds of authentication scenarios. This section seeks to show examples of how the protocol would be applied for different situations.¶
Some longer fields, particularly cryptographic information, have been truncated for display purposes in these examples.¶
In this scenario, the user is the RO and has access to a web browser, and the client can take front-channel callbacks on the same device as the user. This combination is analogous to the OAuth 2 Authorization Code grant type.¶
The client initiates the request to the AS. Here the client identifies itself using its public key.¶
POST /tx HTTP/1.1 Host: server.example.com Content-type: application/json Detached-JWS: ejy0... { "resources": [ { "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] } ], "client": { "key": { "proof": "jwsd", "jwk": { "kty": "RSA", "e": "AQAB", "kid": "xyz-1", "alg": "RS256", "n": "kOB5rR4Jv0GMeLaY6_It_r3ORwdf8ci_JtffXyaSx8xY..." } } }, "interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.example.net/return/123455", "nonce": "LKLTI25DK82FX4T4QFZC" } } }¶
The AS processes the request and determines that the RO needs to interact. The AS returns the following response giving the client the information it needs to connect. The AS has also indicated to the client that it can use the given instance identifier to identify itself in future requests (Section 2.3.1).¶
Content-type: application/json { "interact": { "redirect": "https://server.example.com/interact/4CF492MLVMSW9MKMXKHQ", "callback": "MBDOFXG4Y5CVJCX821LH" } "continue": { "access_token": { "value": "80UPRY5NM33OMUKMKSKU", "key": true }, "uri": "https://server.example.com/continue" }, "instance_id": "7C7C4AZ9KHRS6X63AJAO" }¶
The client saves the response and redirects the user to the interaction_url by sending the following HTTP message to the user's browser.¶
HTTP 302 Found Location: https://server.example.com/interact/4CF492MLVMSW9MKMXKHQ¶
The user's browser fetches the AS's interaction URL. The user logs in, is identified as the RO for the resource being requested, and approves the request. Since the AS has a callback parameter, the AS generates the interaction reference, calculates the hash, and redirects the user back to the client with these additional values added as query parameters.¶
HTTP 302 Found Location: https://client.example.net/return/123455 ?hash=p28jsq0Y2KK3WS__a42tavNC64ldGTBroywsWxT4md_jZQ1R2HZT8BOWYHcLmObM7XHPAdJzTZMtKBsaraJ64A &interact_ref=4IFWWIKYBC2PQ6U56NL1¶
The client receives this request from the user's browser. The client ensures that this is the same user that was sent out by validating session information and retrieves the stored pending request. The client uses the values in this to validate the hash parameter. The client then calls the continuation URL and presents the handle and interaction reference in the request body. The client signs the request as above.¶
POST /continue HTTP/1.1 Host: server.example.com Content-type: application/json Authorization: GNAP 80UPRY5NM33OMUKMKSKU Detached-JWS: ejy0... { "interact_ref": "4IFWWIKYBC2PQ6U56NL1" }¶
The AS retrieves the pending request based on the handle and issues a bearer access token and returns this to the client.¶
Content-type: application/json { "access_token": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": false, "manage": "https://server.example.com/token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L", "resources": [{ "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }] }, "continue": { "access_token": { "value": "80UPRY5NM33OMUKMKSKU", "key": true }, "uri": "https://server.example.com/continue" } }¶
In this scenario, the user does not have access to a web browser on the device and must use a secondary device to interact with the AS. The client can display a user code or a printable QR code. The client prefers a short URL if one is available, with a maximum of 255 characters in length. The is not able to accept callbacks from the AS and needs to poll for updates while waiting for the user to authorize the request.¶
The client initiates the request to the AS.¶
POST /tx HTTP/1.1 Host: server.example.com Content-type: application/json Detached-JWS: ejy0... { "resources": [ "dolphin-metadata", "some other thing" ], "client": "7C7C4AZ9KHRS6X63AJAO", "interact": { "redirect": 255, "user_code": true } }¶
The AS processes this and determines that the RO needs to interact. The AS supports both long and short redirect URIs for interaction, so it includes both. Since there is no "callback" the AS does not include a nonce, but does include a "wait" parameter on the continuation section because it expects the client to poll for results.¶
Content-type: application/json { "interact": { "redirect": "https://srv.ex/MXKHQ", "user_code": { "code": "A1BC-3DFF", "url": "https://srv.ex/device" } }, "continue": { "uri": "https://server.example.com/continue/80UPRY5NM33OMUKMKSKU", "wait": 60 } }¶
The client saves the response and displays the user code visually on its screen along with the static device URL. The client also displays the short interaction URL as a QR code to be scanned.¶
If the user scans the code, they are taken to the interaction endpoint and the AS looks up the current pending request based on the incoming URL. If the user instead goes to the static page and enters the code manually, the AS looks up the current pending request based on the value of the user code. In both cases, the user logs in, is identified as the RO for the resource being requested, and approves the request. Once the request has been approved, the AS displays to the user a message to return to their device.¶
Meanwhile, the client periodically polls the AS every 60 seconds at the continuation URL. The client signs the request using the same key and method that it did in the first request.¶
POST /continue/80UPRY5NM33OMUKMKSKU HTTP/1.1 Host: server.example.com Detached-JWS: ejy0...¶
The AS retrieves the pending request based on the handle and determines that it has not yet been authorized. The AS indicates to the client that no access token has yet been issued but it can continue to call after another 60 second timeout.¶
Content-type: application/json { "continue": { "uri": "https://server.example.com/continue/BI9QNW6V9W3XFJK4R02D", "wait": 60 } }¶
Note that the continuation URL has been rotated since it was used by the client to make this call. The client polls the continuation URL after a 60 second timeout using the new handle.¶
POST /continue/BI9QNW6V9W3XFJK4R02D HTTP/1.1 Host: server.example.com Authorization: GNAP Detached-JWS: ejy0...¶
The AS retrieves the pending request based on the URL, determines that it has been approved, and issues an access token.¶
Content-type: application/json { "access_token": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": false, "manage": "https://server.example.com/token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L", "resources": [ "dolphin-metadata", "some other thing" ] } }¶
In this scenario, the client is requesting access on its own behalf, with no user to interact with.¶
The client creates a request to the AS, identifying itself with its public key and using MTLS to make the request.¶
POST /tx HTTP/1.1 Host: server.example.com Content-type: application/json { "resources": [ "backend service", "nightly-routine-3" ], "client": { "key": { "proof": "mtls", "cert#S256": "bwcK0esc3ACC3DB2Y5_lESsXE8o9ltc05O89jdN-dg2" } } }¶
The AS processes this and determines that the client can ask for the requested resources and issues an access token.¶
Content-type: application/json { "access_token": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": true, "manage": "https://server.example.com/token", "resources": [ "backend service", "nightly-routine-3" ] } }¶
In this scenario, the client is requesting on behalf of a specific RO, but has no way to interact with the user. The AS can asynchronously reach out to the RO for approval in this scenario.¶
The client starts the request at the AS by requesting a set of resources. The client also identifies a particular user.¶
POST /tx HTTP/1.1 Host: server.example.com Content-type: application/json Detached-JWS: ejy0... { "resources": [ { "type": "photo-api", "actions": [ "read", "write", "dolphin" ], "locations": [ "https://server.example.net/", "https://resource.local/other" ], "datatypes": [ "metadata", "images" ] }, "read", "dolphin-metadata", { "type": "financial-transaction", "actions": [ "withdraw" ], "identifier": "account-14-32-32-3", "currency": "USD" }, "some other thing" ], "client": "7C7C4AZ9KHRS6X63AJAO", "user": { "sub_ids": [ { "subject_type": "email", "email": "user@example.com" } ] } }¶
The AS processes this and determines that the RO needs to interact. The AS determines that it can reach the identified user asynchronously and that the identified user does have the ability to approve this request. The AS indicates to the client that it can poll for continuation.¶
Content-type: application/json { "continue": { "access_token": { "value": "80UPRY5NM33OMUKMKSKU", "key": true }, "uri": "https://server.example.com/continue", "wait": 60 } }¶
The AS reaches out to the RO and prompts them for consent. In this example, the AS has an application that it can push notifications in to for the specified account.¶
Meanwhile, the client periodically polls the AS every 60 seconds at the continuation URL.¶
POST /continue HTTP/1.1 Host: server.example.com Authorization: GNAP 80UPRY5NM33OMUKMKSKU Detached-JWS: ejy0...¶
The AS retrieves the pending request based on the handle and determines that it has not yet been authorized. The AS indicates to the client that no access token has yet been issued but it can continue to call after another 60 second timeout.¶
Content-type: application/json { "continue": { "access_token": { "value": "BI9QNW6V9W3XFJK4R02D", "key": true }, "uri": "https://server.example.com/continue", "wait": 60 } }¶
Note that the continuation handle has been rotated since it was used by the client to make this call. The client polls the continuation URL after a 60 second timeout using the new handle.¶
POST /continue HTTP/1.1 Host: server.example.com Authorization: GNAP BI9QNW6V9W3XFJK4R02D Detached-JWS: ejy0...¶
The AS retrieves the pending request based on the handle and determines that it has been approved and it issues an access token.¶
Content-type: application/json { "access_token": { "value": "OS9M2PMHKUR64TB8N6BW7OZB8CDFONP219RP1LT0", "key": false, "manage": "https://server.example.com/token/PRY5NM33OM4TB8N6BW7OZB8CDFONP219RP1L", "resources": [ "dolphin-metadata", "some other thing" ] } }¶
While the GNAP protocol is not designed to be directly compatible with OAuth 2 [RFC6749], considerations have been made to enable the use of OAuth 2 concepts and constructs more smoothly within the GNAP protocol.¶
In this scenario, the client developer has a client_id
and set of
scope
values from their OAuth 2 system and wants to apply them to the
new protocol. Traditionally, the OAuth 2 client developer would put
their client_id
and scope
values as parameters into a redirect request
to the authorization endpoint.¶
HTTP 302 Found Location: https://server.example.com/authorize ?client_id=7C7C4AZ9KHRS6X63AJAO &scope=read%20write%20dolphin &redirect_uri=https://client.example.net/return &response_type=code &state=123455¶
Now the developer wants to make an analogous request to the AS using the new protocol. To do so, the client makes an HTTP POST and places the OAuth 2 values in the appropriate places.¶
POST /tx HTTP/1.1 Host: server.example.com Content-type: application/json Detached-JWS: ejy0... { "resources": [ "read", "write", "dolphin" ], "client": "7C7C4AZ9KHRS6X63AJAO", "interact": { "redirect": true, "callback": { "method": "redirect", "uri": "https://client.example.net/return?state=123455", "nonce": "LKLTI25DK82FX4T4QFZC" } } }¶
The client_id can be used to identify the client's keys that it
uses for authentication, the scopes represent resources that the
client is requesting, and the redirect_uri
and state
value are
pre-combined into a callback
URI that can be unique per request. The
client additionally creates a nonce to protect the callback, separate
from the state parameter that it has added to its return URL.¶
From here, the protocol continues as above.¶
The GNAP protocol makes use of polymorphism within the JSON [RFC8259] structures used for the protocol. Each portion of this protocol is defined in terms of the JSON data type that its values can take, whether it's a string, object, array, boolean, or number. For some fields, different data types offer different descriptive capabilities and are used in different situations for the same field. Each data type provides a different syntax to express the same underlying semantic protocol element, which allows for optimization and simplification in many common cases.¶
Even though JSON is often used to describe strongly typed structures, JSON on its own is naturally polymorphic. In JSON, the named members of an object have no type associated with them, and any data type can be used as the value for any member. In practice, each member has a semantic type that needs to make sense to the parties creating and consuming the object. Within this protocol, each object member is defined in terms of its semantic content, and this semantic content might have expressions in different concrete data types for different specific purposes. Since each object member has exactly one value in JSON, each data type for an object member field is naturally mutually exclusive with other data types within a single JSON object.¶
For example, a resource request for a single access token is composed of an array of resource request descriptions while a request for multiple access tokens is composed of an object whose member values are all arrays. Both of these represent requests for access, but the difference in syntax allows the RC and AS to differentiate between the two request types in the same request.¶
Another form of polymorphism in JSON comes from the fact that the values within JSON arrays need not all be of the same JSON data type. However, within this protocol, each element within the array needs to be of the same kind of semantic element for the collection to make sense, even when the data types are different from each other.¶
For example, each aspect of a resource request can be described using an object with multiple dimensional components, or the aspect can be requested using a string. In both cases, the resource request is being described in a way that the AS needs to interpret, but with different levels of specificity and complexity for the RC to deal with. An API designer can provide a set of common access scopes as simple strings but still allow RC developers to specify custom access when needed for more complex APIs.¶
Extensions to this specification can use different data types for defined fields, but each extension needs to not only declare what the data type means, but also provide justification for the data type representing the same basic kind of thing it extends. For example, an extension declaring an "array" representation for a field would need to explain how the array represents something akin to the non-array element that it is replacing.¶