| RFC 9449 | OAuth DPoP | September 2023 |
| Fett, et al. | Standards Track | [Page] |
This document describes a mechanism for sender-constraining OAuth 2.0tokens via a proof-of-possession mechanism on the application level.This mechanism allows for the detection of replay attacks with access and refreshtokens.¶
This is an Internet Standards Track document.¶
This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 7841.¶
Information about the current status of this document, any errata, and how to provide feedback on it may be obtained athttps://www.rfc-editor.org/info/rfc9449.¶
Copyright (c) 2023 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 Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
Demonstrating Proof of Possession (DPoP) is an application-level mechanism forsender-constraining OAuth[RFC6749] access and refresh tokens. It enables a client toprove the possession of a public/private key pair by includingaDPoP header in an HTTP request. The value of the header is a JSON Web Token(JWT)[RFC7519] that enables the authorizationserver to bind issued tokens to the public part of a client'skey pair. Recipients of such tokens are then able to verify the binding of thetoken to the key pair that the client has demonstrated that it holds viatheDPoP header, thereby providing some assurance that the client presentingthe token also possesses the private key.In other words, the legitimate presenter of the token is constrained to bethe sender that holds and proves possession of the private part of thekey pair.¶
The mechanism specified herein can be used in cases where othermethods of sender-constraining tokens that utilize elements of the underlyingsecure transport layer, such as[RFC8705] or[TOKEN-BINDING],are not available or desirable. For example, due to a sub-par user experienceof TLS client authentication in user agents and a lack of support for HTTP tokenbinding, neither mechanism can be used if an OAuth client is an application thatis dynamically downloaded and executed in a web browser (sometimes referred to as a"single-page application"). Additionally, applications that are installed and run directly on a user's device are well positioned to benefit from DPoP-bound tokens that guard against the misuse of tokens by a compromised or malicious resource.Such applications often have dedicated protected storagefor cryptographic keys.¶
DPoP can be used to sender-constrain access tokens regardless of theclient authentication method employed, but DPoP itself is not used for client authentication.DPoP can also be used to sender-constrain refresh tokens issued to public clients(those without authentication credentials associated with theclient_id).¶
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 specification uses the Augmented Backus-Naur Form (ABNF) notationof[RFC5234].¶
This specification uses the terms "access token", "refresh token","authorization server", "resource server", "authorization endpoint","authorization request", "authorization response", "token endpoint","grant type", "access token request", "access token response","client", "public client", and "confidential client" defined by "The OAuth 2.0 Authorization Framework"[RFC6749].¶
The terms "request", "response", "header field", and "target URI"are imported from[RFC9110].¶
The terms "JOSE" and "JOSE Header" are imported from[RFC7515].¶
This document contains non-normative examples of partial and complete HTTP messages.Some examples use a single trailing backslash to indicate line wrapping for long values, as per[RFC8792].The character and leading spaces on wrapped lines are not part of the value.¶
The primary aim of DPoP is to prevent unauthorized or illegitimateparties from using leaked or stolen access tokens, by binding a tokento a public key upon issuance and requiring that the client provespossession of the corresponding private key when using the token.This constrains the legitimate sender of the token to only the party withaccess to the private key and gives the server receiving the token addedassurances that the sender is legitimately authorized to use it.¶
Access tokens that are sender-constrained via DPoP thus stand incontrast to the typical bearer token, which can be used by any party inpossession of such a token. Although protections generally exist to prevent unintended disclosure of bearer tokens, unforeseen vectors for leakage have occurred due to vulnerabilities and implementation issues in other layers in the protocol or software stack (see, e.g., Compression Ratio Info-leak Made Easy (CRIME)[CRIME], Browser Reconnaissance and Exfiltration via Adaptive Compression of Hypertext (BREACH)[BREACH],Heartbleed[Heartbleed], and the Cloudflare parser bug[Cloudbleed]).There have also been numerous published token theft attacks on OAuthimplementations themselves ([GitHub.Tokens] is just one high-profile example).DPoP provides a general defense in depthagainst the impact of unanticipated token leakage. DPoP is not, however,a substitute for a secure transport andMUST always be used inconjunction with HTTPS.¶
The very nature of the typical OAuth protocol interactionnecessitates that the client discloses the access token to theprotected resources that it accesses. The attacker modelin[SECURITY-TOPICS] describes cases where aprotected resource might be counterfeit, malicious, or compromisedand plays received tokens against other protected resources to gainunauthorized access. Audience-restricted access tokens(e.g., using the JWT[RFC7519]aud claim) canprevent such misuse. However, doing so in practice has proven to beprohibitively cumbersome for many deployments (despite extensions such as[RFC8707]).Sender-constraining access tokens is a more robust and straightforwardmechanism to prevent such token replay at a different endpoint, and DPoPis an accessible application-layer means of doing so.¶
Due to the potential for cross-site scripting (XSS), browser-basedOAuth clients bring to bear added considerations with respect to protectingtokens. The most straightforward XSS-based attack is for an attacker toexfiltrate a token and use it themselves completely independent of thelegitimate client. A stolen access token is used for protectedresource access, and a stolen refresh token is used for obtaining new access tokens.If the private key is non-extractable (as is possible with[W3C.WebCryptoAPI]),DPoP renders exfiltrated tokens alone unusable.¶
XSS vulnerabilities also allow an attacker to execute code in the context ofthe browser-based client application and maliciously use a token indirectlythrough the client. That execution context has access to utilize the signingkey; thus, it can produce DPoP proofs to use in conjunction with the token.At this application layer, there is most likely no feasible defense againstthis threat except generally preventing XSS; therefore, it is consideredout of scope for DPoP.¶
Malicious XSS code executed in the context of the browser-based clientapplication is also in a position to create DPoP proofs with timestamp values inthe future and exfiltrate them in conjunction with a token. These stolenartifacts can later be used independent of the client application toaccess protected resources. To prevent this, servers can optionally requireclients to include a server-chosen value into the proof that cannot be predictedby an attacker (nonce). In the absence of the optional nonce, the impact ofpre-computed DPoP proofs is limited somewhat by the proof being bound to anaccess token on protected resource access. Because a proof covering an accesstoken that does not yet exist cannot feasibly be created, access tokens obtainedwith an exfiltrated refresh token and pre-computed proofs will be unusable.¶
Additional security considerations are discussed inSection 11.¶
The main data structure introduced by this specification is a DPoPproof JWT that is sent as a header in anHTTP request, as described in detail below. A client uses a DPoP proof JWT to provethe possession of a private key corresponding to a certain public key.¶
Roughly speaking, a DPoP proof is a signature over:¶
+--------+ +---------------+| |--(A)-- Token Request ------------------->| || Client | (DPoP Proof) | Authorization || | | Server || |<-(B)-- DPoP-Bound Access Token ----------| || | (token_type=DPoP) +---------------+| || || | +---------------+| |--(C)-- DPoP-Bound Access Token --------->| || | (DPoP Proof) | Resource || | | Server || |<-(D)-- Protected Resource ---------------| || | +---------------++--------+
The basic steps of an OAuth flow with DPoP (without the optional nonce) are shown inFigure 1.¶
The DPoP mechanism presented herein is not a client authentication method.In fact, a primary use case of DPoP is for public clients (e.g., single-pageapplications and applications on a user's device) that do not use client authentication. Nonetheless, DPoPis designed to be compatible withprivate_key_jwt and allother client authentication methods.¶
DPoP does not directly ensure message integrity, but it relies on the TLSlayer for that purpose. SeeSection 11 for details.¶
DPoP introduces the concept of a DPoP proof, which is a JWT created bythe client and sent with an HTTP request using theDPoP header field.Each HTTP request requires a unique DPoP proof.¶
A valid DPoP proof demonstrates to the server that the client holds the privatekey that was used to sign the DPoP proof JWT. This enables authorization servers to bindissued tokens to the corresponding public key (as described inSection 5)and enables resource servers to verify the key-binding of tokens thatit receives (seeSection 7.1), which prevents said tokens frombeing used by any entity that does not have access to the private key.¶
The DPoP proof demonstrates possession of a key and, by itself, is notan authentication or access control mechanism. When presentedin conjunction with a key-bound access token as described inSection 7.1,the DPoP proof provides additional assurance about the legitimacy of the clientto present the access token. However, a valid DPoP proof JWT is not sufficient aloneto make access control decisions.¶
A DPoP proof is included in an HTTP request using the following request header field.¶
DPoP:Figure 2 shows an example DPoP HTTP header field. The example uses "\" line wrapping per[RFC8792].¶
DPoP: eyJ0eXAiOiJkcG9wK2p3dCIsImFsZyI6IkVTMjU2IiwiandrIjp7Imt0eSI6Ik\ VDIiwieCI6Imw4dEZyaHgtMzR0VjNoUklDUkRZOXpDa0RscEJoRjQyVVFVZldWQVdCR\ nMiLCJ5IjoiOVZFNGpmX09rX282NHpiVFRsY3VOSmFqSG10NnY5VERWclUwQ2R2R1JE\ QSIsImNydiI6IlAtMjU2In19.eyJqdGkiOiItQndDM0VTYzZhY2MybFRjIiwiaHRtIj\ oiUE9TVCIsImh0dSI6Imh0dHBzOi8vc2VydmVyLmV4YW1wbGUuY29tL3Rva2VuIiwia\ WF0IjoxNTYyMjYyNjE2fQ.2-GxA6T8lP4vfrg8v-FdWP0A0zdrj8igiMLvqRMUvwnQg\ 4PtFLbdLXiOSsX0x7NVY-FNyJK70nfbV37xRZT3Lg
DPoP HeaderNote that per[RFC9110], header field names are case insensitive; thus,DPoP,DPOP,dpop, etc., are all valid and equivalent headerfield names. However, case is significant in the header field value.¶
TheDPoP HTTP header field valueuses the token68 syntax defined inSection 11.2 of [RFC9110] and is repeated below inFigure 3 for ease of reference.¶
DPoP = token68token68 = 1*( ALPHA / DIGIT / "-" / "." / "_" / "~" / "+" / "/" ) *"="
A DPoP proof is a JWT[RFC7519] that is signed (using JSON Web Signature (JWS)[RFC7515]) with a private key chosen by the client (see below). TheJOSE Header of a DPoP JWTMUST contain at least the following parameters:¶
typ:dpop+jwt, which explicitly types the DPoP proof JWT as recommended inSection 3.11 of [RFC8725].¶alg:none or an identifier for a symmetric algorithm (Message Authentication Code (MAC)).¶jwk:The payload of a DPoP proofMUST contain at least the following claims:¶
jti:jti can be used by the server for replaydetection and prevention; seeSection 11.1.¶htm:htu:iat:When the DPoP proof is used in conjunction with the presentation of an access token in protected resource access (seeSection 7), the DPoP proofMUST also contain the following claim:¶
ath:When the authentication server or resource server provides aDPoP-Nonce HTTP headerin a response (see Sections8 and9), the DPoP proofMUST also containthe following claim:¶
nonce:DPoP-Nonce HTTP header.¶A DPoP proofMAY contain other JOSE Header Parameters or claims as defined by extension,profile, or deployment-specific requirements.¶
Figure 4 is a conceptual example showing the decoded content of the DPoPproof inFigure 2. The JSON of the JWT header and payload are shown,but the signature part is omitted. As usual, line breaks and extra spacesare included for formatting and readability.¶
{ "typ":"dpop+jwt", "alg":"ES256", "jwk": { "kty":"EC", "x":"l8tFrhx-34tV3hRICRDY9zCkDlpBhF42UQUfWVAWBFs", "y":"9VE4jf_Ok_o64zbTTlcuNJajHmt6v9TDVrU0CdvGRDA", "crv":"P-256" }}.{ "jti":"-BwC3ESc6acc2lTc", "htm":"POST", "htu":"https://server.example.com/token", "iat":1562262616}DPoP ProofOf the HTTP request, only the HTTP method and URI areincluded in the DPoP JWT; therefore, only these two message partsare covered by the DPoP proof.The idea is to sign just enough of the HTTP data toprovide reasonable proof of possession with respect to the HTTP request.This design approach of using only a minimal subset of the HTTP headerdata is to avoid the substantial difficulties inherent in attempting tonormalize HTTP messages.Nonetheless, DPoP proofs can be extended to contain other information of theHTTP request (see alsoSection 11.7).¶
To validate a DPoP proof, the receiving serverMUST ensure the following:¶
DPoP HTTP request header field.¶typ JOSE Header Parameter has the valuedpop+jwt.¶alg JOSE Header Parameter indicates a registered asymmetric digitalsignature algorithm[IANA.JOSE.ALGS], is notnone, is supported by theapplication, and is acceptable per local policy.¶jwkJOSE Header Parameter.¶jwk JOSE Header Parameter does not contain a private key.¶htm claim matches the HTTP method of the current request.¶htu claim matches the HTTP URI value for the HTTPrequest in which the JWT was received, ignoring any query andfragment parts.¶nonce claim matches the server-provided nonce value.¶iat claim or a server managed timestamp via thenonce claim, is within an acceptable window (seeSection 11.1).¶If presented to a protected resource in conjunction with an access token,¶
To reduce the likelihood of false negatives,serversSHOULD employ syntax-based normalization (Section 6.2.2 of [RFC3986]) and scheme-basednormalization (Section 6.2.3 of [RFC3986]) before comparing thehtu claim.¶
These checks may be performed in any order.¶
To request an access token that is bound to a public key using DPoP, the clientMUSTprovide a valid DPoP proof JWT in aDPoP header when making an access tokenrequest to the authorization server's token endpoint. This is applicable for allaccess token requests regardless of grant type (e.g.,the commonauthorization_code andrefresh_token grant types and extension grantssuch as the JWT authorization grant[RFC7523]). The HTTP request shown inFigure 5 illustrates such an accesstoken request using an authorization code grant with a DPoP proof JWTin theDPoP header.Figure 5 uses "\" line wrapping per[RFC8792].¶
POST /token HTTP/1.1Host: server.example.comContent-Type: application/x-www-form-urlencodedDPoP: eyJ0eXAiOiJkcG9wK2p3dCIsImFsZyI6IkVTMjU2IiwiandrIjp7Imt0eSI6Ik\ VDIiwieCI6Imw4dEZyaHgtMzR0VjNoUklDUkRZOXpDa0RscEJoRjQyVVFVZldWQVdCR\ nMiLCJ5IjoiOVZFNGpmX09rX282NHpiVFRsY3VOSmFqSG10NnY5VERWclUwQ2R2R1JE\ QSIsImNydiI6IlAtMjU2In19.eyJqdGkiOiItQndDM0VTYzZhY2MybFRjIiwiaHRtIj\ oiUE9TVCIsImh0dSI6Imh0dHBzOi8vc2VydmVyLmV4YW1wbGUuY29tL3Rva2VuIiwia\ WF0IjoxNTYyMjYyNjE2fQ.2-GxA6T8lP4vfrg8v-FdWP0A0zdrj8igiMLvqRMUvwnQg\ 4PtFLbdLXiOSsX0x7NVY-FNyJK70nfbV37xRZT3Lggrant_type=authorization_code\&client_id=s6BhdRkqt\&code=SplxlOBeZQQYbYS6WxSbIA&redirect_uri=https%3A%2F%2Fclient%2Eexample%2Ecom%2Fcb\&code_verifier=bEaL42izcC-o-xBk0K2vuJ6U-y1p9r_wW2dFWIWgjz-
TheDPoP HTTP header fieldMUST contain a valid DPoP proof JWT.If the DPoP proof is invalid, the authorization server issues an errorresponse perSection 5.2 of [RFC6749] withinvalid_dpop_proof as thevalue of theerror parameter.¶
To sender-constrain the access token after checking the validity of theDPoP proof, the authorization server associates the issued access token with thepublic key from the DPoP proof, which can be accomplished as described inSection 6.Atoken_type ofDPoPMUST be included in the access tokenresponse to signal to the client that the access token was bound toits DPoP key and can be used as described inSection 7.1.The example response shown inFigure 6 illustrates such aresponse.¶
HTTP/1.1 200 OKContent-Type: application/jsonCache-Control: no-store{ "access_token": "Kz~8mXK1EalYznwH-LC-1fBAo.4Ljp~zsPE_NeO.gxU", "token_type": "DPoP", "expires_in": 2677, "refresh_token": "Q..Zkm29lexi8VnWg2zPW1x-tgGad0Ibc3s3EwM_Ni4-g"}The example response inFigure 6 includes a refresh token that theclient can use to obtain a new access token when the previous one expires.Refreshing an access token is a token request using therefresh_tokengrant type made to the authorization server's token endpoint. As withall access token requests, the client makes it a DPoP request by includinga DPoP proof, as shown inFigure 7.Figure 7 uses "\" line wrapping per[RFC8792].¶
POST /token HTTP/1.1Host: server.example.comContent-Type: application/x-www-form-urlencodedDPoP: eyJ0eXAiOiJkcG9wK2p3dCIsImFsZyI6IkVTMjU2IiwiandrIjp7Imt0eSI6Ik\ VDIiwieCI6Imw4dEZyaHgtMzR0VjNoUklDUkRZOXpDa0RscEJoRjQyVVFVZldWQVdCR\ nMiLCJ5IjoiOVZFNGpmX09rX282NHpiVFRsY3VOSmFqSG10NnY5VERWclUwQ2R2R1JE\ QSIsImNydiI6IlAtMjU2In19.eyJqdGkiOiItQndDM0VTYzZhY2MybFRjIiwiaHRtIj\ oiUE9TVCIsImh0dSI6Imh0dHBzOi8vc2VydmVyLmV4YW1wbGUuY29tL3Rva2VuIiwia\ WF0IjoxNTYyMjY1Mjk2fQ.pAqut2IRDm_De6PR93SYmGBPXpwrAk90e8cP2hjiaG5Qs\ GSuKDYW7_X620BxqhvYC8ynrrvZLTk41mSRroapUAgrant_type=refresh_token\&client_id=s6BhdRkqt\&refresh_token=Q..Zkm29lexi8VnWg2zPW1x-tgGad0Ibc3s3EwM_Ni4-g
When an authorization server supporting DPoP issues arefresh token to a public client that presents a valid DPoP proof at thetoken endpoint, the refresh tokenMUST be boundto the respective public key. The bindingMUST be validated when the refreshtoken is later presented to get new access tokens. As a result, such a clientMUST present a DPoP proof for the same key that was used to obtain the refreshtoken each time that refresh token is used to obtain a new access token.The implementation details of the binding of the refresh token are at the discretion ofthe authorization server. Since the authorization server both produces andvalidates its refresh tokens, there is no interoperabilityconsideration in the specific details of the binding.¶
An authorization serverMAY elect to issue access tokens that are not DPoP bound,which is signaled to the client with a value ofBearer in thetoken_type parameterof the access token response per[RFC6750]. For a public client that isalso issued a refresh token, this has the effect of DPoP-binding the refresh tokenalone, which can improve the security posture even when protected resources are notupdated to support DPoP.¶
If the access token response contains a differenttoken_type value thanDPoP, theaccess token protection provided by DPoP is not given. The clientMUST discard the response in thiscase if this protection is deemed important for the security of theapplication; otherwise, the client may continue as in a regular OAuth interaction.¶
Refresh tokens issued to confidential clients (those havingestablished authentication credentials with the authorization server)are not bound to the DPoP proof public key because they are alreadysender-constrained with a different existing mechanism. The OAuth 2.0 AuthorizationFramework[RFC6749] already requires that an authorization server bindrefresh tokens to the client to which they were issued and thatconfidential clients authenticate to the authorization server whenpresenting a refresh token. As a result, such refresh tokensare sender-constrained by way of the client identifier and the associatedauthentication requirement. This existing sender-constraining mechanismis more flexible (e.g., it allows credential rotation for the clientwithout invalidating refresh tokens) than binding directly to a particular public key.¶
This document introduces the following authorization server metadata[RFC8414] parameter to signal support for DPoP in general and the specificJWSalg values the authorization server supports for DPoP proof JWTs.¶
dpop_signing_alg_values_supported:alg values (from the[IANA.JOSE.ALGS] registry) supportedby the authorization server for DPoP proof JWTs.¶The Dynamic Client Registration Protocol[RFC7591] defines an APIfor dynamically registering OAuth 2.0 client metadata with authorization servers.The metadata defined by[RFC7591], and registered extensions to it,also imply a general data model for clients that is useful for authorization server implementationseven when the Dynamic Client Registration Protocol isn't in play.Such implementations will typically have some sort of user interface available for managing client configuration.¶
This document introduces the following client registration metadata[RFC7591] parameter to indicate that the client always usesDPoP when requesting tokens from the authorization server.¶
dpop_bound_access_tokens:false.¶If the value istrue, the authorization serverMUST reject token requests from the client that do not contain the DPoP header.¶
Resource serversMUST be able to reliably identify whetheran access token is DPoP-bound and ascertain sufficient informationto verify the binding to the public key of the DPoP proof (seeSection 7.1).Such a binding is accomplished by associating the public keywith the token in a way that can beaccessed by the protected resource, such as embedding the JWKhash in the issued access token directly, using the syntax describedinSection 6.1, or through token introspection as described inSection 6.2. Other methods of associating apublic key with an access token are possible per an agreement by theauthorization server and the protected resource; however, they are beyond thescope of this specification.¶
Resource servers supporting DPoPMUST ensure that the public key fromthe DPoP proof matches the one bound to the access token.¶
When access tokens are represented as JWTs[RFC7519],the public key information is representedusing thejkt confirmation method member defined herein.To convey the hash of a public key in a JWT, this specificationintroduces the following JWT Confirmation Method[RFC7800] member foruse under thecnf claim.¶
jkt:jkt memberMUST be the base64url encoding (as defined in[RFC7515])of the JWK SHA-256 Thumbprint (according to[RFC7638]) of the DPoP public key(in JWK format) to which the access token is bound.¶The following example JWT inFigure 8 with a decoded JWT payload shown inFigure 9 contains acnf claim with thejkt JWK Thumbprint confirmationmethod member. Thejkt value in these examples is the hash of the public keyfrom the DPoP proofs in the examples shown inSection 5.The example uses "\" line wrapping per[RFC8792].¶
eyJhbGciOiJFUzI1NiIsImtpZCI6IkJlQUxrYiJ9.eyJzdWIiOiJzb21lb25lQGV4YW1\wbGUuY29tIiwiaXNzIjoiaHR0cHM6Ly9zZXJ2ZXIuZXhhbXBsZS5jb20iLCJuYmYiOjE\1NjIyNjI2MTEsImV4cCI6MTU2MjI2NjIxNiwiY25mIjp7ImprdCI6IjBaY09DT1JaTll\5LURXcHFxMzBqWnlKR0hUTjBkMkhnbEJWM3VpZ3VBNEkifX0.3Tyo8VTcn6u_PboUmAO\YUY1kfAavomW_YwYMkmRNizLJoQzWy2fCo79Zi5yObpIzjWb5xW4OGld7ESZrh0fsrA
{ "sub":"someone@example.com", "iss":"https://server.example.com", "nbf":1562262611, "exp":1562266216, "cnf": { "jkt":"0ZcOCORZNYy-DWpqq30jZyJGHTN0d2HglBV3uiguA4I" }}"OAuth 2.0 Token Introspection"[RFC7662] defines a method for aprotected resource to query an authorization server about the activestate of an access token. The protected resource also determines metainformationabout the token.¶
For a DPoP-bound access token, the hash of the public key to which the tokenis bound is conveyed to the protected resource as metainformation in a tokenintrospection response. The hash is conveyed using the samecnf content withjkt member structure as the JWK Thumbprint confirmation method, described inSection 6.1, as a top-level member of theintrospection response JSON. Note that the resource serverdoes not send a DPoP proof with the introspection request, and the authorizationserver does not validate an access token's DPoP binding at the introspectionendpoint. Rather, the resource server uses the data of the introspection responseto validate the access token binding itself locally.¶
If thetoken_type member is included in the introspection response, itMUST containthe valueDPoP.¶
The example introspection request inFigure 10 and corresponding response inFigure 11 illustrate an introspection exchange for the example DPoP-boundaccess token that was issued inFigure 6.¶
POST /as/introspect.oauth2 HTTP/1.1Host: server.example.comContent-Type: application/x-www-form-urlencodedAuthorization: Basic cnM6cnM6TWt1LTZnX2xDektJZHo0ZnNON2tZY3lhK1Rptoken=Kz~8mXK1EalYznwH-LC-1fBAo.4Ljp~zsPE_NeO.gxU
HTTP/1.1 200 OKContent-Type: application/jsonCache-Control: no-store{ "active": true, "sub": "someone@example.com", "iss": "https://server.example.com", "nbf": 1562262611, "exp": 1562266216, "cnf": { "jkt": "0ZcOCORZNYy-DWpqq30jZyJGHTN0d2HglBV3uiguA4I" }}Requests to DPoP-protected resourcesMUST include both a DPoP proof as perSection 4 andthe access token as described inSection 7.1.The DPoP proofMUST include theath claim with a valid hash of theassociated access token.¶
Binding the token value to the proof in this way prevents a proofto be used with multiple different access token values across different requests.For example, if a client holds tokens bound to two different resource owners, AT1 and AT2,and uses the same key when talking to the authorization server, it's possible that these tokens could be swapped.Without theath field to bind it, a captured signature applied to AT1 could bereplayed with AT2 instead, changing the rights and access of the intended request.This same substitution prevention remains for rotated access tokens within the samecombination of client and resource owner -- a rotated token value would require thecalculation of a new proof. This binding additionally ensures that a proof intended for usewith the access token is not usable without an access token, or vice-versa.¶
The resource server is required to calculate the hash of the token value presentedand verify that it is the same as the hash value in theath field as described inSection 4.3.Since theath field value is covered by the DPoP proof's signature, its inclusion bindsthe access token value to the holder of the key used to generate the signature.¶
Note that theath field alone does not prevent replay of the DPoP proof or provide bindingto the request in which the proof is presented, and it is still important to check the timewindow of the proof as well as the included message parameters, such ashtm andhtu.¶
A DPoP-bound access token is sent using theAuthorization requestheader field perSection 11.6.2 of [RFC9110] with an authentication scheme ofDPoP. The syntax of theAuthorizationheader field for theDPoP schemeuses the token68 syntax defined inSection 11.2 of [RFC9110] for credentials and is repeated below for ease of reference.The ABNF notation syntax for DPoP authentication scheme credentials is as follows:¶
token68 = 1*( ALPHA / DIGIT / "-" / "." / "_" / "~" / "+" / "/" ) *"="credentials = "DPoP" 1*SP token68
For such an access token, a resource serverMUST check that a DPoP proofwas also received in theDPoP header field of the HTTP request,check the DPoP proof according to the rules inSection 4.3,and check that the public key of the DPoP proof matches the publickey to which the access token is bound perSection 6.¶
The resource serverMUST NOT grant access to the resource unless allchecks are successful.¶
Figure 13 shows an example request to a protectedresource with a DPoP-bound access token in theAuthorization headerand the DPoP proof in theDPoP header.The example uses "\" line wrapping per[RFC8792].Figure 14 shows the decoded content of that DPoPproof. The JSON of the JWT header and payload are shown,but the signature part is omitted. As usual, line breaks and indentationare included for formatting and readability.¶
GET /protectedresource HTTP/1.1Host: resource.example.orgAuthorization: DPoP Kz~8mXK1EalYznwH-LC-1fBAo.4Ljp~zsPE_NeO.gxUDPoP: eyJ0eXAiOiJkcG9wK2p3dCIsImFsZyI6IkVTMjU2IiwiandrIjp7Imt0eSI6Ik\ VDIiwieCI6Imw4dEZyaHgtMzR0VjNoUklDUkRZOXpDa0RscEJoRjQyVVFVZldWQVdCR\ nMiLCJ5IjoiOVZFNGpmX09rX282NHpiVFRsY3VOSmFqSG10NnY5VERWclUwQ2R2R1JE\ QSIsImNydiI6IlAtMjU2In19.eyJqdGkiOiJlMWozVl9iS2ljOC1MQUVCIiwiaHRtIj\ oiR0VUIiwiaHR1IjoiaHR0cHM6Ly9yZXNvdXJjZS5leGFtcGxlLm9yZy9wcm90ZWN0Z\ WRyZXNvdXJjZSIsImlhdCI6MTU2MjI2MjYxOCwiYXRoIjoiZlVIeU8ycjJaM0RaNTNF\ c05yV0JiMHhXWG9hTnk1OUlpS0NBcWtzbVFFbyJ9.2oW9RP35yRqzhrtNP86L-Ey71E\ OptxRimPPToA1plemAgR6pxHF8y6-yqyVnmcw6Fy1dqd-jfxSYoMxhAJpLjA
{ "typ":"dpop+jwt", "alg":"ES256", "jwk": { "kty":"EC", "x":"l8tFrhx-34tV3hRICRDY9zCkDlpBhF42UQUfWVAWBFs", "y":"9VE4jf_Ok_o64zbTTlcuNJajHmt6v9TDVrU0CdvGRDA", "crv":"P-256" }}.{ "jti":"e1j3V_bKic8-LAEB", "htm":"GET", "htu":"https://resource.example.org/protectedresource", "iat":1562262618, "ath":"fUHyO2r2Z3DZ53EsNrWBb0xWXoaNy59IiKCAqksmQEo"}DPoP Proof JWT in Figure 13Upon receipt of a request to a protected resource within the protectionspace requiring DPoP authentication, the server can respond with a challengeto the client to provide DPoP authentication information if the request doesnot include valid credentials or does not contain an access token sufficientfor access. Such a challenge is made using the 401 (Unauthorized) responsestatus code ([RFC9110],Section 15.5.2) and theWWW-Authenticate header field([RFC9110],Section 11.6.1). TheserverMAY include theWWW-Authenticate header inresponse to other conditions as well.¶
In such challenges:¶
DPoP.¶realmMAY be included to indicate thescope of protection in the manner described in[RFC9110],Section 11.5.¶scope authentication parameterMAY be included as defined in[RFC6750],Section 3.¶error parameter ([RFC6750],Section 3)SHOULD be includedto indicate the reason why the request was declined,if the request included an access token but failed authentication.The error parameter values described in[RFC6750],Section 3.1 are suitable,as are any appropriate values defined by extension. The valueuse_dpop_nonce can beused as described inSection 9 to signal that a nonce is needed in the DPoP proof of asubsequent request(s). Additionally,invalid_dpop_proof is used to indicate that the DPoP proofitself was deemed invalid based on the criteria ofSection 4.3.¶error_description parameter ([RFC6750],Section 3)MAY be includedalong with theerror parameter to provide developers a human-readableexplanation that is not meant to be displayed to end-users.¶algs parameterSHOULD be included to signal to the client theJWS algorithms that are acceptable for the DPoP proof JWT.The value of the parameter is a space-delimited list of JWSalg (Algorithm)header values ([RFC7515],Section 4.1.1).¶Figure 15 shows a response to a protected resource request withoutauthentication.¶
HTTP/1.1 401 Unauthorized WWW-Authenticate: DPoP algs="ES256 PS256"
Figure 16 shows a response to a protected resource request that was rejecteddue to the failed confirmation of the DPoP binding in the access token.Figure 16uses "\" line wrapping per[RFC8792].¶
HTTP/1.1 401 UnauthorizedWWW-Authenticate: DPoP error="invalid_token", \ error_description="Invalid DPoP key binding", algs="ES256"
Note that browser-based client applications using Cross-Origin Resource Sharing (CORS)[WHATWG.Fetch] only have accessto CORS-safelisted response HTTP headers by default.In order for the application to obtain and use theWWW-Authenticate HTTP response headervalue, the server needs to make it available to the application by includingWWW-Authenticate in theAccess-Control-Expose-Headers response header list value.¶
This authentication scheme is for origin-server authentication only.Therefore, this authentication schemeMUST NOT be used with theProxy-Authenticate orProxy-Authorization header fields.¶
Note that the syntax of theAuthorization header field for this authentication schemefollows the usage of theBearer scheme defined inSection 2.1 of [RFC6750].While it is not the preferred credential syntax of[RFC9110], it is compatiblewith the general authentication framework therein and is used for consistencyand familiarity with theBearer scheme.¶
Protected resources simultaneously supporting both theDPoP andBearerschemes need to update how the evaluation process is performed for bearer tokens to preventdowngraded usage of a DPoP-bound access token.Specifically, such a protected resourceMUST reject a DPoP-bound accesstoken received as a bearer token per[RFC6750].¶
Section 11.6.1 of [RFC9110] allows a protected resource to indicate support formultiple authentication schemes (i.e.,Bearer andDPoP) with theWWW-Authenticate header field of a 401 (Unauthorized) response.¶
A protected resource that supports only[RFC6750] and is unaware of DPoPwould most presumably accept a DPoP-bound access token as a bearer token(JWT[RFC7519] says to ignore unrecognized claims, Introspection[RFC7662]says that other parameters might be present while placing no functionalrequirements on their presence, and[RFC6750] is effectively silent onthe content of the access token since it relates to validity). As such, aclient can send a DPoP-bound access token using theBearer scheme uponreceipt of aWWW-Authenticate: Bearer challenge from a protected resource(or it can send a DPoP-bound access token if it has prior knowledge of the capabilities of the protectedresource). The effect of this likely simplifies the logistics of phasedupgrades to protected resources in their support DPoP orprolonged deployments of protected resources with mixed token type support.¶
If a protected resource supporting bothBearer andDPoP schemes elects torespond with multipleWWW-Authenticate challenges, attention should be paid towhich challenge(s) should deliver the actual error information. It isRECOMMENDED that the following rules be adhered to:¶
If no authentication information has been included with the request, then thechallengesSHOULD NOT include an error code or other error information, as perSection 3.1 of [RFC6750] (Figure 17).¶
If the mechanism used to attempt authentication could be establishedunambiguously, then the corresponding challengeSHOULD be used to deliver errorinformation (Figure 18).¶
Otherwise, bothBearer andDPoP challengesMAY be used to deliver error information (Figure 19).¶
The following examples use "\" line wrapping per[RFC8792].¶
GET /protectedresource HTTP/1.1Host: resource.example.orgHTTP/1.1 401 UnauthorizedWWW-Authenticate: Bearer, DPoP algs="ES256 PS256"
GET /protectedresource HTTP/1.1Host: resource.example.orgAuthorization: Bearer INVALID_TOKENHTTP/1.1 401 UnauthorizedWWW-Authenticate: Bearer error="invalid_token", \ error_description="Invalid token", DPoP algs="ES256 PS256"
GET /protectedresource HTTP/1.1Host: resource.example.orgAuthorization: Bearer Kz~8mXK1EalYznwH-LC-1fBAo.4Ljp~zsPE_NeO.gxUAuthorization: DPoP Kz~8mXK1EalYznwH-LC-1fBAo.4Ljp~zsPE_NeO.gxUHTTP/1.1 400 Bad RequestWWW-Authenticate: Bearer error="invalid_request", \ error_description="Multiple methods used to include access token", \ DPoP algs="ES256 PS256", error="invalid_request", \ error_description="Multiple methods used to include access token"
Authorization including a DPoP proof may not be idempotent (depending on serverenforcement ofjti,iat, andnonce claims). Consequently, all previouslyidempotent requests for protected resources that were previously idempotent mayno longer be idempotent. It isRECOMMENDED that clients generate a unique DPoPproof, even when retrying idempotent requests in response to HTTP errorsgenerally understood as transient.¶
Clients that encounter frequent network errors may experience additionalchallenges when interacting with servers with stricter nonce validationimplementations.¶
This section specifies a mechanism using opaque nonces provided by the serverthat can be used to limit the lifetime of DPoP proofs.Without employing such a mechanism, a malicious party controlling the client(potentially including the end-user)can create DPoP proofs for use arbitrarily far in the future.¶
Including a nonce value contributed by the authorization server in the DPoP proofMAY be used by authorization servers to limit the lifetime of DPoP proofs. The server determines when to issue a new DPoP nonce challenge and if it is needed, thereby requiring the use of the nonce value in subsequent DPoP proofs.The logic through which the server makes that determination is out of scope of this document.¶
An authorization serverMAY supply a nonce value to be included by the clientin DPoP proofs sent. In this case, the authorization server responds to requests that do not include a noncewith an HTTP 400 (Bad Request) error response perSection 5.2 of [RFC6749] usinguse_dpop_nonce as theerror code value. The authorization server includes aDPoP-Nonce HTTP header in the response supplyinga nonce value to be used when sending the subsequent request. Nonce valuesMUST be unpredictable.This same error code is used when supplying a new nonce value when there was a nonce mismatch.The client will typically retry the request with the new nonce value suppliedupon receiving ause_dpop_nonce error with an accompanying nonce value.¶
For example, in response to a token request without a nonce when the authorization server requires one,the authorization server can respond with aDPoP-Nonce value such as the following to providea nonce value to include in the DPoP proof:¶
HTTP/1.1 400 Bad Request DPoP-Nonce: eyJ7S_zG.eyJH0-Z.HX4w-7v { "error": "use_dpop_nonce", "error_description": "Authorization server requires nonce in DPoP proof" }Other HTTP headers and JSON fieldsMAY also be included in the error response,but thereMUST NOT be more than oneDPoP-Nonce header.¶
Upon receiving the nonce, the client is expected to retry its token requestusing a DPoP proof including the supplied nonce value in thenonce claimof the DPoP proof.An example unencoded JWT payload of such a DPoP proof including a nonce is shown below.¶
{ "jti": "-BwC3ESc6acc2lTc", "htm": "POST", "htu": "https://server.example.com/token", "iat": 1562262616, "nonce": "eyJ7S_zG.eyJH0-Z.HX4w-7v" }The nonce is opaque to the client.¶
If thenonce claim in the DPoP proofdoes not exactly match a nonce recently supplied by the authorization server to the client,the authorization serverMUST reject the request.The rejection responseMAY include aDPoP-Nonce HTTP headerproviding a new nonce value to use for subsequent requests.¶
The intent is that clients need to keep only one nonce value and servers need to keep awindow of recent nonces.That said, transient circumstances may arise in which the stored nonce values for the server and the client differ.However, this situation is self-correcting.With any rejection message,the server can send the client the nonce value it wants to useto the client, and the client can store that nonce value and retry the request with it.Even if the client and/or server discard their stored nonce values,that situation is also self-correcting because new nonce values can be communicatedwhen responding to or retrying failed requests.¶
Note that browser-based client applications using CORS[WHATWG.Fetch] only have accessto CORS-safelisted response HTTP headers by default.In order for the application to obtain and use theDPoP-Nonce HTTP response headervalue, the server needs to make it available to the application by includingDPoP-Nonce in theAccess-Control-Expose-Headers response header list value.¶
The nonce syntax in ABNF as used by[RFC6749](which is the same as the scope-token syntax) is shown below.¶
nonce = 1*NQCHAR
It is up to the authorization server when to supply a new nonce valuefor the client to use.The client is expected to use the existing supplied nonce in DPoP proofsuntil the server supplies a new nonce value.¶
The authorization serverMAY supply the new nonce in the same way thatthe initial one was supplied: by using aDPoP-Nonce HTTP header in the response.TheDPoP-Nonce HTTP header field uses the nonce syntax defined inSection 8.1.Each time this happens, it requires an extra protocol round trip.¶
A more efficient manner of supplying a new nonce value is also definedby including aDPoP-Nonce HTTP headerin the HTTP 200 (OK) response from the previous request.The clientMUST use the new nonce value supplied for the next token requestand for all subsequent token requests until the authorization serversupplies a new nonce.¶
Responses that include theDPoP-Nonce HTTP header should be uncacheable(e.g., usingCache-Control: no-store in response to aGET request) toprevent the response from being used to serve a subsequent request and a stalenonce value from being used as a result.¶
An example 200 OK response providing a new nonce value is shown below.¶
HTTP/1.1 200 OK Cache-Control: no-store DPoP-Nonce: eyJ7S_zG.eyJbYu3.xQmBj-1
Resource servers can also choose to provide a nonce value to be includedin DPoP proofs sent to them.They provide the nonce using theDPoP-Nonce header in the same way that authorization servers doas described in Sections8 and8.2.The error signaling is performed as described inSection 7.1.Resource servers use an HTTP 401 (Unauthorized) error codewith an accompanyingWWW-Authenticate: DPoP valueandDPoP-Nonce value to accomplish this.¶
For example, in response to a resource request without a nonce when the resource server requires one,the resource server can respond with aDPoP-Nonce value such as the following to providea nonce value to include in the DPoP proof.The example below uses "\" line wrapping per[RFC8792].¶
HTTP/1.1 401 Unauthorized WWW-Authenticate: DPoP error="use_dpop_nonce", \ error_description="Resource server requires nonce in DPoP proof" DPoP-Nonce: eyJ7S_zG.eyJH0-Z.HX4w-7v
Note that the nonces provided by an authorization server and a resource server are differentand should not be confused with one anothersince nonces will be only accepted by the server that issued them.Likewise, should a client use multiple authorization servers and/or resource servers,a nonce issued by any of them should be used only at the issuing server.Developers should also be careful to not confuse DPoP nonces with theOpenID Connect[OpenID.Core] ID Token nonce.¶
Binding the authorization code issued to the client's proof-of-possession keycan enable end-to-end binding of the entire authorization flow.This specification defines thedpop_jkt authorization request parameter for this purpose.The value of thedpop_jkt authorization request parameter is theJWK Thumbprint[RFC7638] of the proof-of-possession public keyusing the SHA-256 hash function, which isthe same value as used for thejkt confirmation method defined inSection 6.1.¶
When a token request is received, the authorization server computes theJWK Thumbprint of the proof-of-possession public key in the DPoP proofand verifies that it matches thedpop_jkt parameter value in the authorization request.If they do not match, itMUST reject the request.¶
An example authorization request using thedpop_jkt authorization request parameter is shown below and uses "\" line wrapping per[RFC8792].¶
GET /authorize?response_type=code&client_id=s6BhdRkqt3&state=xyz\ &redirect_uri=https%3A%2F%2Fclient%2Eexample%2Ecom%2Fcb\ &code_challenge=E9Melhoa2OwvFrEMTJguCHaoeK1t8URWbuGJSstw-cM\ &code_challenge_method=S256\ &dpop_jkt=NzbLsXh8uDCcd-6MNwXF4W_7noWXFZAfHkxZsRGC9Xs HTTP/1.1Host: server.example.com
dpop_jkt ParameterUse of thedpop_jkt authorization request parameter isOPTIONAL.Note that thedpop_jkt authorization request parameterMAY also be usedin combination with Proof Key for Code Exchange (PKCE)[RFC7636], which is recommended by[SECURITY-TOPICS]as a countermeasure to authorization code injection. Thedpop_jkt authorizationrequest parameter only provides similar protections when a unique DPoP key isused for each authorization request.¶
When Pushed Authorization Requests (PARs)[RFC9126] are used in conjunction with DPoP, there are two ways in which the DPoP key can be communicated in the PAR request:¶
dpop_jkt parameter can be used as described inSection 10 to bind the issuedauthorization code to a specific key. In this case,dpop_jktMUST be included alongside other authorization request parameters in the POST body of the PAR request.¶DPoP header can be added to the PAR request. In thiscase, the authorization serverMUST check the provided DPoP proof JWT asdefined inSection 4.3. ItMUST further behave as if the contained public key'sthumbprint was provided usingdpop_jkt, i.e., reject the subsequent tokenrequest unless a DPoP proof for the same key is provided. This can help tosimplify the implementation of the client, as it can "blindly" attach theDPoP header to all requests to the authorization server regardless of thetype of request. Additionally, it provides a stronger binding, as theDPoPheader contains a proof of possession of the private key.¶Both mechanismsMUST be supported by an authorization server that supports PAR and DPoP. If both mechanisms are used at the same time, the authorization serverMUST reject the request if the JWK Thumbprint indpop_jkt does not match the public key in theDPoP header.¶
Allowing both mechanisms ensures that clients usingdpop_jkt do not need todistinguish between front-channel and pushed authorization requests, and at thesame time, clients that only have one code path for protecting all calls to authorization serverendpoints do not need to distinguish between requests to the PAR endpoint and thetoken endpoint.¶
In DPoP, the prevention of token replay at a different endpoint (seeSection 2) is achieved through authentication of the server per[RFC6125] andthe binding of the DPoP proof to a certain URI and HTTP method. However, DPoPhas a somewhat different nature of protection than TLS-basedmethods such as OAuth Mutual TLS[RFC8705] or OAuth TokenBinding[TOKEN-BINDING] (see also Sections11.1 and11.7).TLS-based mechanisms can leverage a tight integrationbetween the TLS layer and the application layer to achieve strongmessage integrity, authenticity,and replay protection.¶
If an adversary is able to get hold of a DPoP proof JWT, the adversarycould replay that token at the same endpoint (the HTTP endpointand method are enforced via the respective claims in the JWTs). Tolimit this, serversMUST only accept DPoP proofs for a limited timeafter their creation (preferably only for a relatively brief periodon the order of seconds or minutes).¶
In the context of the target URI, servers can store thejti value ofeach DPoP proof for the time window in which the respective DPoP proof JWT would be accepted to prevent multiple uses of the same DPoP proof.HTTP requests to the same URI for which thejti value has been seen beforewould be declined. When strictly enforced, such a single-use check provides a very strong protection against DPoPproof replay, but it may not always be feasible in practice, e.g., whenmultiple servers behind a single endpoint have no shared state.¶
In order to guard againstmemory exhaustion attacks, a server that is trackingjti values should rejectDPoP proof JWTs with unnecessarily largejti values or store only a hash thereof.¶
Note: To accommodate for clock offsets, the serverMAY accept DPoPproofs that carry aniat time in the reasonably near future (on the order of seconds or minutes).Because clock skews between serversand clients may be large, serversMAY limit DPoP proof lifetimes by usingserver-provided nonce values containing the time at the server rather thancomparing the client-suppliediat time to the time at the server. Noncescreated in this way yield the same result even in the face of arbitrarilylarge clock skews.¶
Server-provided nonces are an effective means for further reducing the chances for successful DPoP proof replay.Unlike cryptographic nonces, it is acceptable for clients to use the samenonce multiple times and for the server to accept the same nonce multipletimes. As long as thejti value is tracked and duplicates are rejected for the lifetime of thenonce, thereis no additional risk of token replay.¶
An attacker in control of the client can pre-generate DPoP proofs forspecific endpoints arbitrarily far into the future by choosing theiat value in the DPoP proof to be signed by the proof-of-possession key.Note that one such attacker is the person who is the legitimate user of the client.The user may pre-generate DPoP proofs to exfiltratefrom the machine possessing the proof-of-possession keyupon which they were generatedand copy them to another machine that does not possess the key.For instance, a bank employee might pre-generate DPoP proofson a bank computer and then copy them to another machinefor use in the future, thereby bypassing bank audit controls.When DPoP proofs can be pre-generated and exfiltrated,all that is actually being proved in DPoP protocol interactionsis possession of a DPoP proof -- not of the proof-of-possession key.¶
Use of server-provided nonce values that are not predictable by attackers can prevent this attack.By providing new nonce values at times of its choosing,the server can limit the lifetime of DPoP proofs,preventing pre-generated DPoP proofs from being used.When server-provided nonces are used, possessionof the proof-of-possession key is being demonstrated --not just possession of a DPoP proof.¶
Theath claim limits the use of pre-generated DPoP proofs to the lifetimeof the access token. Deployments that do not utilize the nonce mechanismSHOULD NOT issue long-lived DPoP constrained access tokens,preferring instead to use short-lived access tokens and refresh tokens.Whilst an attacker could pre-generate DPoP proofs to use the refresh tokento obtain a new access token, they would be unable to realisticallypre-generate DPoP proofs to use a newly issued access token.¶
A serverMUST NOT accept any DPoP proofs without thenonce claim when a DPoP nonce has been provided to the client.¶
If an adversary is able to run code in the client's execution context,the security of DPoP is no longer guaranteed. Common issues in webapplications leading to the execution of untrusted code are XSS and remote code inclusion attacks.¶
If the private key used for DPoP is stored in such a way that itcannot be exported, e.g., in a hardware or software security module,the adversary cannot exfiltrate the key and use it to create arbitraryDPoP proofs. The adversary can, however, create new DPoP proofs aslong as the client is online and uses these proofs (together with therespective tokens) either on the victim's device or on a device underthe attacker's control to send arbitrary requests that will beaccepted by servers.¶
To send requests even when the client is offline, an adversary can tryto pre-compute DPoP proofs using timestamps in the future andexfiltrate these together with the access or refresh token.¶
An adversary might further try to associate tokens issued from thetoken endpoint with a key pair under the adversary's control. One wayto achieve this is to modify existing code, e.g., by replacingcryptographic APIs. Another way is to launch a new authorization grantbetween the client and the authorization server in an iframe. Thisgrant needs to be "silent", i.e., not require interaction with theuser. With code running in the client's origin, the adversary hasaccess to the resulting authorization code and can use it to associatetheir own DPoP keys with the tokens returned from the token endpoint.The adversary is then able to use the resulting tokens on their owndevice even if the client is offline.¶
Therefore, protecting clients against the execution of untrusted codeis extremely important even if DPoP is used. Besides secure codingpractices, Content Security Policy[W3C.CSP] can be used as a secondlayer of defense against XSS.¶
Servers accepting signed DPoP proof JWTsMUST verify that thetyp field isdpop+jwt in theheaders of the JWTs to ensure that adversaries cannot use JWTs createdfor other purposes.¶
ImplementersMUST ensure that only asymmetric digital signature algorithms (such asES256) thatare deemed secure can be used for signing DPoP proofs. In particular,the algorithmnoneMUST NOT be allowed.¶
DPoP does not ensure the integrity of the payload or headers ofrequests. The DPoP proof only contains claims for the HTTP URI andmethod, but not the message body or general requestheaders, for example.¶
This is an intentional design decision intended to keep DPoP simple to use, butas described, it makes DPoP potentially susceptible to replay attackswhere an attacker is able to modify message contents and headers. Inmany setups, the message integrity and confidentiality provided by TLSis sufficient to provide a good level of protection.¶
Note: While signatures covering other parts of requests are out of the scope ofthis specification, additional information to be signed can beadded into DPoP proofs.¶
The binding of the access token to the DPoP public key, as specified inSection 6, uses a cryptographic hash of the JWKrepresentation of the public key. It relieson the hash function having sufficient second-preimage resistance soas to make it computationally infeasible to find or create anotherkey that produces to the same hash output value. The SHA-256hash function was used because it meets the aforementionedrequirement while being widely available.¶
Similarly, the binding of the DPoP proof to the access token uses ahash of that access token as the value of theath claimin the DPoP proof (seeSection 4.2). This relies on the valueof the hash being sufficiently unique so as to reliably identify theaccess token. The collision resistance of SHA-256 meets that requirement.¶
Cryptographic binding of the authorization code to the DPoP public keyis specified inSection 10. This binding prevents attacks in which the attacker captures the authorization code and creates a DPoP proof using a proof-of-possession key other than the one held by the client and redeems the authorization code using that DPoP proof. By ensuring end to end that only the client's DPoP key can be used, this prevents captured authorization codes from being exfiltrated and used at locations other than the one to which the authorization code was issued.¶
Authorization codes can, for instance, be harvested by attackersfrom places where the HTTP messages containing them are logged.Even when efforts are made to make authorization codes one-time-use, in practice,there is often a time window during which attackers can replay them.For instance, when authorization servers are implemented as scalable replicated services,some replicas may temporarily not yet have the information needed to prevent replay.DPoP binding of the authorization code solves these problems.¶
If an authorization server does not (or cannot) strictly enforce the single-use limitation for authorization codesand an attacker can access the authorization code (and if PKCE is used, thecode_verifier),the attacker can create a forged token request, binding the resulting token to an attacker-controlled key.For example, using XSS, attackers might obtain access to the authorization code and PKCE parameters.Use of thedpop_jkt parameter prevents this attack.¶
The binding of the authorization code to the DPoP public keyuses a JWK Thumbprint of the public key, just as the access token binding does.The same JWK Thumbprint considerations apply.¶
Thejkt confirmation method member, theath JWT claim, and thedpop_jkt authorizationrequest parameter defined herein all use the output of the SHA-256 hash function as their value.The use of a single hash function by this specification was intentional and aimed atsimplicity and avoidance of potential security and interoperability issues arising fromcommon mistakes implementing and deploying parameterized algorithm agility schemes.However, the use of a different hash function is not precluded if future circumstanceschange and make SHA-256 insufficient for the requirements of this specification.Should that need arise, it is expected that a short specification will be produced thatupdates this one. Using the output of an appropriatehash function as the value, that specification will likely define a new confirmation method member, a new JWT claim,and a new authorization request parameter. These items will be used in place of, or alongside, theirrespective counterparts in the same message structures and flows of the larger protocol definedby this specification.¶
In cases where DPoP is used with client authentication, it is only bound to authentication by beingcoincident in the same TLS tunnel. Since the DPoP proof is not directly boundto the authentication cryptographically, it's possible that the authentication or the DPoP messages were copied intothe tunnel. While including the URI in the DPoP can partially mitigate some of this risk, modifyingthe authentication mechanism to provide cryptographic binding between authentication and DPoP couldprovide better protection. However, providing additional binding with authentication through themodification of authentication mechanisms or other means is beyond the scope of this specification.¶
IANA has registered the following access tokentype in the "OAuth Access Token Types" registry[IANA.OAuth.Params]established by[RFC6749].¶
IANA has registered the following error valuesin the "OAuth Extensions Error" registry[IANA.OAuth.Params]established by[RFC6749].¶
IANA has registered the following authorization request parameterin the "OAuth Parameters" registry[IANA.OAuth.Params]established by[RFC6749].¶
dpop_jkt¶IANA has registered the following scheme in the"HTTP Authentication Schemes" registry[IANA.HTTP.AuthSchemes] established by[RFC9110],Section 16.4.1.¶
DPoP¶IANA has registered theapplication/dpop+jwt media type[RFC2046]in the "Media Types" registry[IANA.MediaTypes] in the manner described in[RFC6838],which is used to indicate that the content is a DPoP JWT.¶
IANA has registered the following JWTcnf member valuein the "JWT Confirmation Methods" registry[IANA.JWT]established by[RFC7800].¶
IANA has registered the following Claims in the "JSON Web Token Claims" registry[IANA.JWT] established by[RFC7519].¶
htm¶htu¶ath¶The Internet Security Glossary[RFC4949] provides a useful definition of nonceas a random or non-repeating value that is included in dataexchanged by a protocol, usually for the purpose of guaranteeingliveness and thus detecting and protecting against replay attacks.¶
However, the initial registration of thenonce claim by[OpenID.Core]used language that was contextually specific to that application,which was potentially limiting to its general applicability.¶
Therefore, IANA has updated the entry fornonce in the"JSON Web Token Claims" registry[IANA.JWT] with an expanded definition to reflectthat the claim can be used appropriately in other contexts and with the addition of this document as a reference, as follows.¶
nonce¶IANA has registered the following HTTP header fields, as specified by this document, in the "Hypertext Transfer Protocol (HTTP) Field Name Registry"[IANA.HTTP.Fields] established by[RFC9110]:¶
IANA has registered the following valuein the "OAuth Authorization Server Metadata" registry[IANA.OAuth.Params]established by[RFC8414].¶
dpop_signing_alg_values_supported¶IANA has registered the following valuein the IANA "OAuth Dynamic Client Registration Metadata" registry[IANA.OAuth.Params]established by[RFC7591].¶
dpop_bound_access_tokens¶We would like to thankBrock Allen,Annabelle Backman,Dominick Baier,Spencer Balogh,Vittorio Bertocci,Jeff Corrigan,Domingos Creado,Philippe De Ryck,Andrii Deinega,William Denniss,Vladimir Dzhuvinov,Mike Engan,Nikos Fotiou,Mark Haine,Dick Hardt,Joseph Heenan,Bjorn Hjelm,Jacob Ideskog,Jared Jennings,Benjamin Kaduk,Pieter Kasselman,Neil Madden,Rohan Mahy,Karsten Meyer zu Selhausen,Nicolas Mora,Steinar Noem,Mark Nottingham,Rob Otto,Aaron Parecki,Michael Peck,Roberto Polli,Paul Querna,Justin Richer,Joseph Salowey,Rifaat Shekh-Yusef,Filip Skokan,Dmitry Telegin,Dave Tonge,Jim Willeke,and others for their valuable input, feedback, and general support of this work.¶
This document originated from discussions at the 4th OAuth SecurityWorkshop in Stuttgart, Germany. We thank the organizers of thisworkshop (Ralf Küsters andGuido Schmitz).¶