RFC 8705: OAuth 2.0 Mutual-TLS Client Authentication and Certificate-Bound Access Tokens

Status of This Memo

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 at https://www.rfc-editor.org/info/rfc8705.¶

Copyright Notice

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.¶

1. Introduction

The OAuth 2.0 Authorization Framework [RFC6749] enables third-party client applications to obtain delegated access to protected resources. In the prototypical abstract OAuth flow, illustrated in Figure 1, the client obtains an access token from an entity known as an authorization server and then uses that token when accessing protected resources, such as HTTPS APIs.¶

  +--------+                                 +---------------+
  |        |                                 |               |
  |        |<--(A)-- Get an access token --->| Authorization |
  |        |                                 |     Server    |
  |        |                                 |               |
  |        |                                 +---------------+
  |        |                                         ^
  |        |                                         |
  |        |
  |        |                               (C)       |
  | Client |                           Validate the
  |        |                           access token  |
  |        |
  |        |                                         |
  |        |                                         v
  |        |                                 +---------------+
  |        |                                 |      (C)      |
  |        |                                 |               |
  |        |<--(B)-- Use the access token -->|   Protected   |
  |        |                                 |    Resource   |
  |        |                                 |               |
  +--------+                                 +---------------+

The flow illustrated in Figure 1 includes the following steps:¶

(A)

The client makes an HTTPS POST request to the authorization server and presents a credential representing the authorization grant. For certain types of clients (those that have been issued or otherwise established a set of client credentials) the request must be authenticated. In the response, the authorization server issues an access token to the client.¶

(B)

The client includes the access token when making a request to access a protected resource.¶

(C)

The protected resource validates the access token in order to authorize the request. In some cases, such as when the token is self-contained and cryptographically secured, the validation can be done locally by the protected resource. Other cases require that the protected resource call out to the authorization server to determine the state of the token and obtain metainformation about it.¶

Layering on the abstract flow above, this document standardizes enhanced security options for OAuth 2.0 utilizing client-certificate-based mutual TLS. Section 2 provides options for authenticating the request in Step (A). Step (C) is supported with semantics to express the binding of the token to the client certificate for both local and remote processing in Sections 3.1 and 3.2, respectively. This ensures that, as described in Section 3, protected resource access in Step (B) is only possible by the legitimate client using a certificate-bound token and holding the private key corresponding to the certificate.¶

OAuth 2.0 defines a shared-secret method of client authentication but also allows for defining and using additional client authentication mechanisms when interacting directly with the authorization server. This document describes an additional mechanism of client authentication utilizing mutual-TLS certificate-based authentication that provides better security characteristics than shared secrets. While [RFC6749] documents client authentication for requests to the token endpoint, extensions to OAuth 2.0 (such as Introspection [RFC7662], Revocation [RFC7009], and the Backchannel Authentication Endpoint in [OpenID.CIBA]) define endpoints that also utilize client authentication, and the mutual-TLS methods defined herein are applicable to those endpoints as well.¶

Mutual-TLS certificate-bound access tokens ensure that only the party in possession of the private key corresponding to the certificate can utilize the token to access the associated resources. Such a constraint is sometimes referred to as key confirmation, proof-of-possession, or holder-of-key and is unlike the case of the bearer token described in [RFC6750], where any party in possession of the access token can use it to access the associated resources. Binding an access token to the client's certificate prevents the use of stolen access tokens or replay of access tokens by unauthorized parties.¶

Mutual-TLS certificate-bound access tokens and mutual-TLS client authentication are distinct mechanisms that are complementary but don't necessarily need to be deployed or used together.¶

Additional client metadata parameters are introduced by this document in support of certificate-bound access tokens and mutual-TLS client authentication. The authorization server can obtain client metadata via the Dynamic Client Registration Protocol [RFC7591], which defines mechanisms for dynamically registering OAuth 2.0 client metadata with authorization servers. Also the metadata defined by [RFC7591], and registered extensions to it, imply a general data model for clients that is useful for authorization server implementations, even 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.¶

2. Mutual TLS for OAuth Client Authentication

This section defines, as an extension of Section 2.3 of OAuth 2.0 [RFC6749], two distinct methods of using mutual-TLS X.509 client certificates as client credentials. The requirement of mutual TLS for client authentication is determined by the authorization server, based on policy or configuration for the given client (regardless of whether the client was dynamically registered, statically configured, or otherwise established).¶

In order to utilize TLS for OAuth client authentication, the TLS connection between the client and the authorization server MUST have been established or re-established with mutual-TLS X.509 certificate authentication (i.e., the client Certificate and CertificateVerify messages are sent during the TLS handshake).¶

For all requests to the authorization server utilizing mutual-TLS client authentication, the client MUST include the client_id parameter described in Section 2.2 of OAuth 2.0 [RFC6749]. The presence of the client_id parameter enables the authorization server to easily identify the client independently from the content of the certificate. The authorization server can locate the client configuration using the client identifier and check the certificate presented in the TLS handshake against the expected credentials for that client. The authorization server MUST enforce the binding between client and certificate, as described in either Section 2.1 or 2.2 below. If no certificate is presented, or that which is presented doesn't match that which is expected for the given client_id, the authorization server returns a normal OAuth 2.0 error response per Section 5.2 of [RFC6749] with the invalid_client error code to indicate failed client authentication.¶

3. Mutual-TLS Client Certificate-Bound Access Tokens

When mutual TLS is used by the client on the connection to the token endpoint, the authorization server is able to bind the issued access token to the client certificate. Such a binding is accomplished by associating the certificate with the token in a way that can be accessed by the protected resource, such as embedding the certificate hash in the issued access token directly, using the syntax described in Section 3.1, or through token introspection as described in Section 3.2. Binding the access token to the client certificate in that fashion has the benefit of decoupling that binding from the client's authentication with the authorization server, which enables mutual TLS during protected resource access to serve purely as a proof-of-possession mechanism. Other methods of associating a certificate with an access token are possible, per agreement by the authorization server and the protected resource, but are beyond the scope of this specification.¶

In order for a resource server to use certificate-bound access tokens, it must have advance knowledge that mutual TLS is to be used for some or all resource accesses. In particular, the access token itself cannot be used as input to the decision of whether or not to request mutual TLS because (from the TLS perspective) it is "Application Data", only exchanged after the TLS handshake has been completed, and the initial CertificateRequest occurs during the handshake, before the Application Data is available. Although subsequent opportunities for a TLS client to present a certificate may be available, e.g., via TLS 1.2 renegotiation [RFC5246] or TLS 1.3 post-handshake authentication [RFC8446], this document makes no provision for their usage. It is expected to be common that a mutual-TLS-using resource server will require mutual TLS for all resources hosted thereupon or will serve mutual-TLS-protected and regular resources on separate hostname and port combinations, though other workflows are possible. How resource server policy is synchronized with the authorization server (AS) is out of scope for this document.¶

Within the scope of a mutual-TLS-protected resource-access flow, the client makes protected resource requests, as described in [RFC6750], however, those requests MUST be made over a mutually authenticated TLS connection using the same certificate that was used for mutual TLS at the token endpoint.¶

The protected resource MUST obtain, from its TLS implementation layer, the client certificate used for mutual TLS and MUST verify that the certificate matches the certificate associated with the access token. If they do not match, the resource access attempt MUST be rejected with an error, per [RFC6750], using an HTTP 401 status code and the invalid_token error code.¶

Metadata to convey server and client capabilities for mutual-TLS client certificate-bound access tokens is defined in Sections 3.3 and 3.4, respectively.¶

  {
    "iss": "https://server.example.com",
    "sub": "ty.webb@example.com",
    "exp": 1493726400,
    "nbf": 1493722800,
    "cnf":{
      "x5t#S256": "bwcK0esc3ACC3DB2Y5_lESsXE8o9ltc05O89jdN-dg2"
    }
  }

  HTTP/1.1 200 OK
  Content-Type: application/json

  {
    "active": true,
    "iss": "https://server.example.com",
    "sub": "ty.webb@example.com",
    "exp": 1493726400,
    "nbf": 1493722800,
    "cnf":{
      "x5t#S256": "bwcK0esc3ACC3DB2Y5_lESsXE8o9ltc05O89jdN-dg2"
    }
  }

4. Public Clients and Certificate-Bound Tokens

Mutual-TLS OAuth client authentication and certificate-bound access tokens can be used independently of each other. Use of certificate-bound access tokens without mutual-TLS OAuth client authentication, for example, is possible in support of binding access tokens to a TLS client certificate for public clients (those without authentication credentials associated with the client_id). The authorization server would configure the TLS stack in the same manner as for the Self-Signed Certificate method such that it does not verify that the certificate presented by the client during the handshake is signed by a trusted CA. Individual instances of a client would create a self-signed certificate for mutual TLS with both the authorization server and resource server. The authorization server would not use the mutual-TLS certificate to authenticate the client at the OAuth layer but would bind the issued access token to the certificate for which the client has proven possession of the corresponding private key. The access token is then bound to the certificate and can only be used by the client possessing the certificate and corresponding private key and utilizing them to negotiate mutual TLS on connections to the resource server. When the authorization server issues a refresh token to such a client, it SHOULD also bind the refresh token to the respective certificate and check the binding when the refresh token is presented to get new access tokens. The implementation details of the binding of the refresh token are at the discretion of the authorization server.¶

5. Metadata for Mutual-TLS Endpoint Aliases

The process of negotiating client certificate-based mutual TLS involves a TLS server requesting a certificate from the TLS client (the client does not provide one unsolicited). Although a server can be configured such that client certificates are optional, meaning that the connection is allowed to continue when the client does not provide a certificate, the act of a server requesting a certificate can result in undesirable behavior from some clients. This is particularly true of web browsers as TLS clients, which will typically present the end user with an intrusive certificate selection interface when the server requests a certificate.¶

Authorization servers supporting both clients using mutual TLS and conventional clients MAY chose to isolate the server side mutual-TLS behavior to only clients intending to do mutual TLS, thus avoiding any undesirable effects it might have on conventional clients. The following authorization server metadata parameter is introduced to facilitate such separation:¶

mtls_endpoint_aliases

OPTIONAL. A JSON object containing alternative authorization server endpoints that, when present, an OAuth client intending to do mutual TLS uses in preference to the conventional endpoints. The parameter value itself consists of one or more endpoint parameters, such as token_endpoint, revocation_endpoint, introspection_endpoint, etc., conventionally defined for the top level of authorization server metadata. An OAuth client intending to do mutual TLS (for OAuth client authentication and/or to acquire or use certificate-bound tokens) when making a request directly to the authorization server MUST use the alias URL of the endpoint within the mtls_endpoint_aliases, when present, in preference to the endpoint URL of the same name at the top level of metadata. When an endpoint is not present in mtls_endpoint_aliases, then the client uses the conventional endpoint URL defined at the top level of the authorization server metadata. Metadata parameters within mtls_endpoint_aliases that do not define endpoints to which an OAuth client makes a direct request have no meaning and SHOULD be ignored.¶

Below is an example of an authorization server metadata document with the mtls_endpoint_aliases parameter, which indicates aliases for the token, revocation, and introspection endpoints that an OAuth client intending to do mutual TLS would use in preference to the conventional token, revocation, and introspection endpoints. Note that the endpoints in mtls_endpoint_aliases use a different host than their conventional counterparts, which allows the authorization server (via TLS server_name extension [RFC6066] or actual distinct hosts) to differentiate its TLS behavior as appropriate.¶

{
  "issuer": "https://server.example.com",
  "authorization_endpoint": "https://server.example.com/authz",
  "token_endpoint": "https://server.example.com/token",
  "introspection_endpoint": "https://server.example.com/introspect",
  "revocation_endpoint": "https://server.example.com/revo",
  "jwks_uri": "https://server.example.com/jwks",
  "response_types_supported": ["code"],
  "response_modes_supported": ["fragment","query","form_post"],
  "grant_types_supported": ["authorization_code", "refresh_token"],
  "token_endpoint_auth_methods_supported":
                  ["tls_client_auth","client_secret_basic","none"],
  "tls_client_certificate_bound_access_tokens": true,
  "mtls_endpoint_aliases": {
    "token_endpoint": "https://mtls.example.com/token",
    "revocation_endpoint": "https://mtls.example.com/revo",
    "introspection_endpoint": "https://mtls.example.com/introspect"
  }
}

6. Implementation Considerations

7. Security Considerations

8. Privacy Considerations

In TLS versions prior to 1.3, the client's certificate is sent unencrypted in the initial handshake and can potentially be used by third parties to monitor, track, and correlate client activity. This is likely of little concern for clients that act on behalf of a significant number of end users because individual user activity will not be discernible amidst the client activity as a whole. However, clients that act on behalf of a single end user, such as a native application on a mobile device, should use TLS version 1.3 whenever possible or consider the potential privacy implications of using mutual TLS on earlier versions.¶

9. IANA Considerations

Appendix A. Example "cnf" Claim, Certificate, and JWK

For reference, an x5t#S256 value and the X.509 certificate from which it was calculated are provided in the following examples, Figures 5 and 6, respectively. A JWK representation of the certificate's public key along with the x5c member is also provided in Figure 7.¶

"cnf":{"x5t#S256":"A4DtL2JmUMhAsvJj5tKyn64SqzmuXbMrJa0n761y5v0"}

-----BEGIN CERTIFICATE-----
MIIBBjCBrAIBAjAKBggqhkjOPQQDAjAPMQ0wCwYDVQQDDARtdGxzMB4XDTE4MTAx
ODEyMzcwOVoXDTIyMDUwMjEyMzcwOVowDzENMAsGA1UEAwwEbXRsczBZMBMGByqG
SM49AgEGCCqGSM49AwEHA0IABNcnyxwqV6hY8QnhxxzFQ03C7HKW9OylMbnQZjjJ
/Au08/coZwxS7LfA4vOLS9WuneIXhbGGWvsDSb0tH6IxLm8wCgYIKoZIzj0EAwID
SQAwRgIhAP0RC1E+vwJD/D1AGHGzuri+hlV/PpQEKTWUVeORWz83AiEA5x2eXZOV
bUlJSGQgjwD5vaUaKlLR50Q2DmFfQj1L+SY=
-----END CERTIFICATE-----

{
 "kty":"EC",
 "x":"1yfLHCpXqFjxCeHHHMVDTcLscpb07KUxudBmOMn8C7Q",
 "y":"8_coZwxS7LfA4vOLS9WuneIXhbGGWvsDSb0tH6IxLm8",
 "crv":"P-256",
 "x5c":[
  "MIIBBjCBrAIBAjAKBggqhkjOPQQDAjAPMQ0wCwYDVQQDDARtdGxzMB4XDTE4MTA
   xODEyMzcwOVoXDTIyMDUwMjEyMzcwOVowDzENMAsGA1UEAwwEbXRsczBZMBMGBy
   qGSM49AgEGCCqGSM49AwEHA0IABNcnyxwqV6hY8QnhxxzFQ03C7HKW9OylMbnQZ
   jjJ/Au08/coZwxS7LfA4vOLS9WuneIXhbGGWvsDSb0tH6IxLm8wCgYIKoZIzj0E
   AwIDSQAwRgIhAP0RC1E+vwJD/D1AGHGzuri+hlV/PpQEKTWUVeORWz83AiEA5x2
   eXZOVbUlJSGQgjwD5vaUaKlLR50Q2DmFfQj1L+SY="
   ]
 }

Appendix B. Relationship to Token Binding

OAuth 2.0 Token Binding [TOKEN] enables the application of Token Binding to the various artifacts and tokens employed throughout OAuth. That includes binding of an access token to a Token Binding key, which bears some similarities in motivation and design to the mutual-TLS client certificate-bound access tokens defined in this document. Both documents define what is often called a proof-of-possession security mechanism for access tokens, whereby a client must demonstrate possession of cryptographic keying material when accessing a protected resource. The details differ somewhat between the two documents but both have the authorization server bind the access token that it issues to an asymmetric key pair held by the client. The client then proves possession of the private key from that pair with respect to the TLS connection over which the protected resource is accessed.¶

Token Binding uses bare keys that are generated on the client, which avoids many of the difficulties of creating, distributing, and managing certificates used in this specification. However, at the time of writing, Token Binding is fairly new, and there is relatively little support for it in available application development platforms and tooling. Until better support for the underlying core Token Binding specifications exists, practical implementations of OAuth 2.0 Token Binding are infeasible. Mutual TLS, on the other hand, has been around for some time and enjoys widespread support in web servers and development platforms. As a consequence, OAuth 2.0 Mutual-TLS Client Authentication and Certificate-Bound Access Tokens can be built and deployed now using existing platforms and tools. In the future, the two specifications are likely to be deployed in parallel for solving similar problems in different environments. Authorization servers may even support both specifications simultaneously using different proof-of-possession mechanisms for tokens issued to different clients.¶

Acknowledgements

Scott "not Tomlinson" Tomilson and Matt Peterson were involved in design and development work on a mutual-TLS OAuth client authentication implementation that predates this document. Experience and learning from that work informed some of the content of this document.¶

This specification was developed within the OAuth Working Group under the chairmanship of Hannes Tschofenig and Rifaat Shekh-Yusef with Eric Rescorla, Benjamin Kaduk, and Roman Danyliw serving as Security Area Directors. Additionally, the following individuals contributed ideas, feedback, and wording that helped shape this specification: Vittorio Bertocci, Sergey Beryozkin, Ralph Bragg, Sophie Bremer, Roman Danyliw, Vladimir Dzhuvinov, Samuel Erdtman, Evan Gilman, Leif Johansson, Michael Jones, Phil Hunt, Benjamin Kaduk, Takahiko Kawasaki, Sean Leonard, Kepeng Li, Neil Madden, James Manger, Jim Manico, Nov Matake, Sascha Preibisch, Eric Rescorla, Justin Richer, Vincent Roca, Filip Skokan, Dave Tonge, and Hannes Tschofenig.¶

Authors' Addresses