Network Working Group                                      Paul E. Jones
Internet Draft                                         Gonzalo Salgueiro
Intended status: Standards Track                           Cisco Systems
Expires: June 2, 2013                                       Joseph Smarr
                                                        December 2, 2012



   This specification defines the WebFinger protocol, which can be used
   to discover information about people or other entities on the
   Internet using standard HTTP methods.

Status of this Memo

   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 http://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 June 2, 2013.

Copyright Notice

   Copyright (c) 2012 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
   (http://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.

Jones, et al.            Expires June 2, 2013                   [Page 1]

Internet-Draft                WebFinger                    December 2012

Table of Contents

   1. Introduction...................................................2
   2. Terminology....................................................3
   3. Overview.......................................................3
   4. Example Use of WebFinger.......................................3
      4.1. Locating a User's Blog....................................3
      4.2. Identity Provider Discovery for OpenID Connect............5
      4.3. Auto-Configuration of Email Clients.......................6
      4.4. Retrieving Device Information.............................7
   5. WebFinger Protocol.............................................8
      5.1. Performing a WebFinger Query..............................8
      5.2. The JSON Resource Descriptor (JRD)........................9
         5.2.1. expires..............................................9
         5.2.2. subject.............................................10
         5.2.3. aliases.............................................10
         5.2.4. properties..........................................10
         5.2.5. links...............................................11
      5.3. The "rel" Parameter......................................13
      5.4. WebFinger and URIs.......................................14
   6. Cross-Origin Resource Sharing (CORS)..........................15
   7. Access Control................................................15
   8. Hosted WebFinger Services.....................................16
   9. Security Considerations.......................................17
   10. IANA Considerations..........................................18
   11. Acknowledgments..............................................19
   12. References...................................................19
      12.1. Normative References....................................19
      12.2. Informative References..................................20
   Author's Addresses...............................................20

1. Introduction

   WebFinger is used to discover information about people or other
   entities on the Internet using standard HTTP [2] methods.  The
   WebFinger server returns a JavaScript Object Notation (JSON) [5]
   object that describes the resource being queried.  The JSON object is
   referred to as the JSON Resource Descriptor (JRD).  The JRD contains
   link relations, properties, titles, and other information that is
   suitable for automated processing.  For a person, the kinds of
   information that might be shared via WebFinger include a personal
   profile address, identity service, telephone number, or preferred
   avatar.  For other entities on the Internet, the server might return
   JRDs containing link relations that allow a client to discover the
   amount of toner in a printer or the physical location of a server.

   Information returned via WebFinger might be for direct human
   consumption (e.g., looking up someone's phone number), or it might be

Jones, et al.            Expires June 2, 2013                   [Page 2]

Internet-Draft                WebFinger                    December 2012

   used by systems to help carry out some operation (e.g., facilitate
   logging into a web site by determining a user's identity service).

2. Terminology

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   document are to be interpreted as described in RFC 2119 [1].

   WebFinger makes heavy use of "Link Relations".  Briefly, a Link
   Relation is an attribute and value pair used on the Internet wherein
   the attribute identifies the type of link to which the associated
   value refers.  In Hypertext Transfer Protocol (HTTP) and Web Linking
   [4], the attribute is a "rel" and the value is an "href". WebFinger
   also uses the "rel" attribute, where the "rel" value is either a
   single IANA-registered link relation type [11] or a URI [6].

3. Overview

   WebFinger enables the discovery of information about users, devices,
   and other entities that are associated with a host.  Discovery
   involves a single HTTP GET request to the well-known [3] "webfinger"
   resource at the target host and receiving back a JavaScript Object
   Notation (JSON) [5] Resource Descriptor (JRD) (see section 5.2)
   containing link relations, properties, titles, and other useful
   information.  The request MUST include the URI or IRI [7] for the
   entity for which information is sought as a parameter named

   Use of WebFinger is illustrated in the examples in Section 4, then
   described more formally in Section 5.

4. Example Use of WebFinger

   This non-normative section shows a few sample uses of WebFinger.

4.1. Locating a User's Blog

   Assume you receive an email from Bob and he refers to something he
   posted on his blog, but you do not know where Bob's blog is located.
   It would be simple to discover the address of Bob's blog if he makes
   that information available via WebFinger.

   Assume your email client can discover the blog for you.  After
   receiving the message from Bob (bob@example.com), you instruct your
   email client to perform a WebFinger query.  It does so by issuing the
   following HTTPS query to example.com:

Jones, et al.            Expires June 2, 2013                   [Page 3]

Internet-Draft                WebFinger                    December 2012

     GET /.well-known/webfinger?
                           resource=acct%3Abob%40example.com HTTP/1.1
     Host: example.com

   The server might then respond with a message like this:

     HTTP/1.1 200 OK
     Access-Control-Allow-Origin: *
     Content-Type: application/json; charset=UTF-8

       "expires" : "2012-11-16T19:41:35Z",
       "subject" : "acct:bob@example.com",
       "aliases" :
       "properties" :
           "http://example.com/rel/role/" : "employee"
       "links" :
           "rel" : "http://webfinger.net/rel/avatar",
           "type" : "image/jpeg",
           "href" : "http://www.example.com/~bob/bob.jpg"
           "rel" : "http://webfinger.net/rel/profile-page",
           "href" : "http://www.example.com/~bob/"
           "rel" : "blog",
           "type" : "text/html",
           "href" : "http://blogs.example.com/bob/",
           "titles" :
               "en-us" : "The Magical World of Bob",
               "fr" : "Le monde magique de Bob"
           "rel" : "vcard",
           "href" : "http://www.example.com/~bob/bob.vcf"

Jones, et al.            Expires June 2, 2013                   [Page 4]

Internet-Draft                WebFinger                    December 2012

   The email client would take note of the "blog" link relation in the
   above JRD that refers to Bob's blog.  This URL would then be
   presented to you so that you could then visit his blog.  The email
   client might also note that Bob has published an avatar link relation
   and use that picture to represent Bob inside the email client.
   Lastly, the client might consider the vcard [15] link relation in
   order to update contact information for Bob.

   In the above example, an "acct" URI [8] is used in the query, though
   any valid alias for the user might also be used.  An alias is a URI
   that is different from the "subject" URI that identifies the same
   entity.  In the above example, there is one "http" alias returned,
   though there might have been more than one.  Had the "http:" URI
   shown as an alias been used to query for information about Bob, the
   query would have appeared as:

     GET /.well-known/webfinger?
              resource=http%3A%2F%2Fwww.example.com%2F~bob%2F HTTP/1.1
     Host: example.com

   The response would have been substantially the same, with the subject
   and alias information changed as necessary.  Other information, such
   as the expiration time might also change, but the set of link
   relations and properties would be the same with either response.

4.2. Identity Provider Discovery for OpenID Connect

   Suppose Carol wishes to authenticate with a web site she visits using
   OpenID Connect [17].  She would provide the web site with her OpenID
   Connect identifier, say carol@example.com.  The visited web site
   would perform a WebFinger query looking for the OpenID Connect
   Provider.  Since the site is interested in only one particular link
   relation, the server might utilize the "rel" parameter as described
   in section 5.3:

     GET /.well-known/webfinger?
     Host: example.com

   The server might respond with a JRD like this:

       "expires" : "2012-11-16T19:41:35Z",
       "subject" : "acct:carol@example.com",
       "aliases" :

Jones, et al.            Expires June 2, 2013                   [Page 5]

Internet-Draft                WebFinger                    December 2012

       "properties" :
           "http://example.com/rel/role/" : "employee"
       "links" :
           "rel" : "http://openid.net/specs/connect/1.0/issuer",
           "href" : "https://openid.example.com/"

   Since the "rel" parameter only filters the link relations returned by
   the server, other name/value pairs in the response, including any
   aliases or properties, would be returned.  Also, since support for
   the "rel" parameter is optional, the client must not assume the
   "links" array will contain only the requested link relation.

4.3. Auto-Configuration of Email Clients

   WebFinger could be used to auto-provision an email client with basic
   configuration data.  Suppose that sue@example.com wants to configure
   her email client.  Her email client might issue the following query:

     GET /.well-known/webfinger?
              resource=mailto%3Asue%40example.com HTTP/1.1
     Host: example.com

   The response from the server would contain entries for the various
   protocols, transport options, and security options.  If there are
   multiple options, the server might return a link relation that for
   each of the valid options and the client or Sue might select which
   option to choose.  Since JRDs list link relations in a specific
   order, then the most-preferred choices could be presented first.
   Consider this response:

     "subject" : "mailto:sue@example.com",
     "links" :
         "rel" : "http://example.net/rel/smtp-server",
         "properties" :
           "http://example.net/email/host" : "smtp.example.com",
           "http://example.net/email/port" : "587",
           "http://example.net/email/login-required" : "yes",

Jones, et al.            Expires June 2, 2013                   [Page 6]

Internet-Draft                WebFinger                    December 2012

           "http://example.net/email/transport" : "starttls"
         "rel" : "http://example.net/rel/imap-server",
         "properties" :
           "http://example.net/email/host" : "imap.example.com",
           "http://example.net/email/port" : "993",
           "http://example.net/email/transport" : "ssl"

   In this example, you can see that the WebFinger server advertises an
   SMTP service and an IMAP service.  In this example, the "href"
   entries associated with the link relation are absent.  This is valid
   when there is no external reference that needs to be made.

4.4. Retrieving Device Information

   As another example, suppose there are printers on the network and you
   would like to check the current toner level for a particular printer
   identified via the URI device:p1.example.com.  While the "device" URI
   scheme is not presently specified, we use it here for illustrative

   Following the procedures similar to those above, a query may be
   issued to get link relations specific to this URI like this:

     GET /.well-known/webfinger?resource=
                       device%3Ap1.example.com HTTP/1.1
     Host: example.com

   The link relations that are returned for a device may be quite
   different than those for user accounts.  Perhaps we may see a
   response like this:

     HTTP/1.1 200 OK
     Access-Control-Allow-Origin: *
     Content-Type: application/json; charset=UTF-8

       "subject" : "device:p1.example.com",
       "links" :
           "rel" : "http://example.com/rel/tipsi",

Jones, et al.            Expires June 2, 2013                   [Page 7]

Internet-Draft                WebFinger                    December 2012

           "href" : ""

   While this example is fictitious, you can imagine that perhaps the
   Transport Independent, Printer/System Interface [16] may be enhanced
   with a web interface that allows a device that understands the TIP/SI
   web interface specification to query the printer for toner levels.

5. WebFinger Protocol

   WebFinger is a simple HTTP-based web service that returns a JSON
   Resource Descriptor (JRD) to convey information about an entity on
   the Internet and the Cross-Origin Resource Sharing (CORS) [10]
   specification to facilitate queries made via a web browser.

   This specification defines URI parameters that are passed from the
   client to the server when issuing a request.  These parameters,
   "resource" and "rel", and the parameter values are included in the
   "query" component of the URI (see Section 3.4 of RFC 3986).  To
   construct the "query" component, the client performs the following
   steps.  First, each parameter value is percent-encoded as per Section
   2.1 of RFC 3986.  Next, the client constructs a string to be placed
   in the query component by concatenating the name of the first
   parameter together with an equal sign ("=") and the percent-encoded
   parameter value.  For any subsequent parameters, the client appends
   an ampersand ("&") to the string, the name of the next parameter, an
   equal sign, and percent-encoded parameter value.  The client MUST NOT
   insert any spaces while constructing the string.  The order in which
   the client places each parameter and its corresponding parameter
   value is unspecified.

5.1. Performing a WebFinger Query

   WebFinger clients issue queries to the well-known resource /.well-
   known/webfinger.  All queries MUST include the "resource" parameter
   exactly once and set to the value of the URI for which information is
   being sought.  If the "resource" parameter is absent or malformed,
   the WebFinger server MUST return a 400 status code.

   Clients MUST query the server using HTTPS and utilize HTTP only if an
   HTTPS connection cannot be established, and then only if the client
   issuing the query will not utilize information in the response in
   such a way as to compromise user security or privacy.  As an example,
   a client using WebFinger to facilitate logging into a web site MUST
   only utilize HTTPS to ensure that a user is not misdirected to a
   rogue web site that might steal the user's credentials.  If the HTTPS
   server has an invalid certificate or returns an HTTP status code

Jones, et al.            Expires June 2, 2013                   [Page 8]

Internet-Draft                WebFinger                    December 2012

   indicating some error, including a 4xx or 5xx, the client MUST NOT
   use HTTP in attempt to complete the discovery.

   WebFinger servers MUST return a JRD as the representation for the
   resource if the client requests no format explicitly via the HTTP
   "Accept" header.  A client MAY include the "Accept" header to
   indicate a desired representation, though no other representation is
   defined in this specification.  The media type used for the JSON
   Resource Descriptor (JRD) is "application/json" [5].

   If the client queries the WebFinger server and provides a URI for
   which the server has no information, the server MUST return a 404
   status code.

   WebFinger servers can include cache validators in a response to
   enable conditional requests by clients and/or expiration times as per
   RFC 2616 section 13.

5.2. The JSON Resource Descriptor (JRD)

   The JSON Resource Descriptor (JRD) is a JSON object that is comprised
   of name/value pairs appearing in this RECOMMENDED order:

        o expires
        o subject
        o aliases
        o properties
        o links

   The members "expires" and "subject" are name/value pairs whose value
   are strings, "aliases" is an array of strings, "properties" is an
   object comprised of name/value pairs whose values are strings, and
   "links" is an array of objects that contain link relation

   When processing a JRD, the client MUST ignore any unknown member and
   not treat the presence of an unknown member as an error.

   Below, each of these members of the JRD is described in more detail.

5.2.1. expires

   The value of the "expires" member is a string that indicates the date
   and time after which the JRD SHOULD be considered expired and no
   longer utilized.  The format of the date/time string is:


Jones, et al.            Expires June 2, 2013                   [Page 9]

Internet-Draft                WebFinger                    December 2012

   Here, "YYYY" indicates the four-digit year, "MM" indicates the two-
   digit month (in the range of 01 to 12), and "DD" indicates the two-
   digit day of the month (in the range of 01 to 31).  The "T" is
   literally an ASCII "T" that exists merely as a separator between the
   date and the time.  The "HH" indicates the two-digit hour of the day
   (in the range of 01 to 12), "MM" indicates the two-digit minute of
   the day (in the range of 00 to 59), and "SS" indicates the two-digit
   number of seconds (in the range of 00 to 59).  A colon (":")
   character MUST separate the hours, minutes, and seconds values, and a
   hyphen ("-") MUST separate the year, month, and day in the string.
   The "Z" at the end of the string is literally an ASCII "Z" that
   indicates UTC time and MUST be present.  The "expires" string MUST
   utilize UTC time.  An example of the "expires" member is:

     "expires" : "2012-11-16T19:41:35Z"

   The server MAY include the "expires" header in a JRD and clients
   SHOULD honor the value if present.

5.2.2. subject

   The value of the "subject" member is a string that MUST be set to the
   same value as the "resource" parameter in the client request.  This
   is a URI that identifies the entity for which the client queried the

   The "subject" member MUST be included in the JRD.

5.2.3. aliases

   The "aliases" array is an array of zero or more URI strings that
   identify the same entity as the "subject" URI. Each URI must be an
   absolute URI.

   The server MAY include the "aliases" array in the JRD.

5.2.4. properties

   The "properties" object is comprised of zero or more name/value pairs
   whose names are absolute URIs and whose values are strings or null.
   Properties are used to convey additional information about the
   subject of the JRD.  As an example, consider this use of

     "properties" : { "http://webfinger.net/rel/name" : "Bob Smith" }

   The server MAY include the "properties" member in the JRD.

Jones, et al.            Expires June 2, 2013                  [Page 10]

Internet-Draft                WebFinger                    December 2012

5.2.5. links

   The "links" array contains zero or more elements that contain the
   link relation information.  Each element of the array is an object
   comprised of the following name/value pairs in this RECOMMENDED

        o rel
        o type
        o href
        o titles
        o properties

   The members "rel", "type", and "href" are a name/value pairs whose
   values are strings, "titles" and "properties" are objects comprised
   of name/value pairs whose values are strings.

   The order of elements in the "links" array indicates an order of
   preference.  Thus, if there are two or more link relations having the
   same "rel" value, the first link relation would indicate the user's
   preferred link relation.

   Servers MAY include the "links" array in the JRD.

   Below, each of the members of the objects found in the "links" array
   is described in more detail.  Each object in the "links" array,
   referred to as a "link relation object", is completely independent
   from any other object in the array; any requirement on the server to
   include a given member in the link relation object refers only to
   that particular object. rel

   The value of the "rel" member is a string that is either an absolute
   URI or a registered relation type [11] (see RFC 5988 [4]).  The value
   of the "rel" member MUST contain exactly one URI string or registered
   relation type and MUST NOT contain a space-separated list of URIs or
   registered relation types.  The URI or registered relation type
   identifies the type of the link relation.  The other members of the
   object have meaning only once the type of link relation is
   understood.  In some instances, the link relation will have
   associated semantics that allow a client to query for other resources
   on the Internet.  In other instances, the link relation will have
   associated semantics that allow the client to utilize the other
   members of the link relation object without fetching additional
   external resources.

   Servers MUST include the "rel" member in the link relation object.

Jones, et al.            Expires June 2, 2013                  [Page 11]

Internet-Draft                WebFinger                    December 2012 type

   The value of the "type" member is a string that indicates the media
   type [12] of the linked resource (see RFC 4288 [13]).

   Servers MAY include the "type" member in the link relation object. href

   The value of the "href" member is a string that contains a URI
   pointing to the linked resource.

   Servers MAY include the "href" member in the link relation object.  titles

   The "titles" object is comprised of zero or more name/value pairs
   whose name is a language tag [14] or the string "default".  The
   string is human-readable and describes the link relation.  More than
   one title for the link relation MAY be provided for the benefit of
   users who utilize the link relation and, if used, a language
   identifier SHOULD be duly used as the name.  If the language is
   unknown or unspecified, then the name is "default".

   A server SHOULD NOT include more than one title named with the same
   language tag (or "default") within the link relation object.  The
   client behavior is undefined if a link relation object includes more
   than one title named with the same language tag (or "default"),
   though the client MUST NOT treat this as an error.  The client can
   select whichever title or titles it wishes to utilize.

   Here is an example of the titles object:

     "titles" :
       "en-us" : "The Magical World of Bob",
       "fr" : "Le monde magique de Bob"

   The server MAY include the "titles" member in the link relation
   object. properties

   The "properties" object within the link relation object is comprised
   of zero or more name/value pairs whose names are absolute URIs and
   whose values are strings or null.  Properties are used to convey
   additional information about the link relation.  As an example,
   consider this use of "properties":

Jones, et al.            Expires June 2, 2013                  [Page 12]

Internet-Draft                WebFinger                    December 2012

     "properties" : { "http://example.net/mail/port" : "993" }

   The server MAY include the "properties" member in the link relation

5.3. The "rel" Parameter

   When issuing a request to the server, the client MAY utilize the
   "rel" parameter to request only a subset of the information that
   would otherwise be returned without the "rel" parameter.  When the
   "rel" parameter is used, only the link relations that match the link
   relations provided via "rel" are included in the array of links
   returned in the JRD.  All other information normally present in a
   resource descriptor is present in the resource descriptor, even when
   "rel" is employed.

   The "rel" parameter MAY be transmitted to the server multiple times
   in order to request multiple types of link relations.

   The purpose of the "rel" parameter is to return a subset of
   resource's link relations.  Use of the parameter might reduce
   processing requirements on either the client or server, and it might
   also reduce the bandwidth required to convey the partial resource
   descriptor, especially if there are numerous link relation values to
   convey for a given resource.

   Support for the "rel" parameter is OPTIONAL, but RECOMMENDED on the
   server.  Should the server not support the "rel" parameter, it MUST
   ignore it and process the request as if no "rel" parameter values
   were present.

   The following example presents the same example as found in section
   4.1, but uses the "rel" parameter in order to select two link

     GET /.well-known/webfinger?
                rel=vcard HTTP/1.1
     Host: example.com

   In this example, the client requests the link relations of type
   "http://webfinger.net/rel/profile-page" and "vcard".  The server then
   responds with a message like this:

     HTTP/1.1 200 OK
     Access-Control-Allow-Origin: *
     Content-Type: application/json; charset=UTF-8

Jones, et al.            Expires June 2, 2013                  [Page 13]

Internet-Draft                WebFinger                    December 2012

       "expires" : "2012-11-16T19:41:35Z",
       "subject" : "acct:bob@example.com",
       "aliases" :
       "properties" :
           "http://example.com/rel/role/" : "employee"
       "links" :
           "rel" : "http://webfinger.net/rel/profile-page",
           "href" : "http://www.example.com/~bob/"
           "rel" : "vcard",
           "href" : "http://www.example.com/~bob/bob.vcf"

   As you can see, the server returned only the link relations requested
   by the client, but also included the other parts of the JRD.

   In the event that a client requests links for link relations that are
   not defined for the specified resource, a resource descriptor MUST be
   returned.  In the returned JRD, the "links" array MAY be absent,
   empty, or contain only links that did match a provided "rel" value.
   The server MUST NOT return a 404 status code when a particular link
   relation specified via "rel" is not defined for the resource, as a
   404 status code is reserved for indicating that the resource itself
   (e.g., either /.well-known/webfinger or the resource indicated via
   the "resource" parameter) does not exist.

5.4. WebFinger and URIs

   WebFinger requests can include a parameter specifying the URI of an
   account, device, or other entity.  WebFinger is agnostic regarding
   the scheme of such a URI: it could be an "acct" URI [7], an "http" or
   "https" URI, a "mailto" URI, or some other scheme.

   For resources associated with a user account at a host, use of the
   "acct" URI scheme is RECOMMENDED, since it explicitly identifies an
   account accessible via WebFinger.  Further, the "acct" URI scheme is
   not associated with other protocols as, by way of example, the
   "mailto" URI scheme is associated with email.  Since not every host

Jones, et al.            Expires June 2, 2013                  [Page 14]

Internet-Draft                WebFinger                    December 2012

   offers email service, using the "mailto" URI scheme [9] is not ideal
   for identifying user accounts on all hosts.  That said, use of the
   "mailto" URI scheme would be ideal for use with WebFinger to discover
   mail server configuration information for a user.

   A host MAY utilize one or more URIs that serve as aliases for the
   user's account, such as URIs that use the "http" URI scheme [2].  A
   WebFinger server MUST return substantially the same response to both
   an "acct" URI and any alias URI for the account, including the same
   set of link relations and properties.  The only name/value pairs in
   the response that MAY be different include "subject", "expires", and
   "aliases".  In addition, the server SHOULD include the entire list
   aliases for the user's account in the JRD returned when querying the
   LRDD resource or when utilizing the "resource" parameter.

6. Cross-Origin Resource Sharing (CORS)

   WebFinger resources might not be accessible from a web browser due to
   "Same-Origin" policies.  The current best practice is to make
   resources available to browsers through Cross-Origin Resource Sharing
   (CORS) [10], and servers MUST include the Access-Control-Allow-Origin
   HTTP header in responses.  Servers SHOULD support the least
   restrictive setting by allowing any domain access to the WebFinger

      Access-Control-Allow-Origin: *

   There are cases where defaulting to the least restrictive setting is
   not appropriate, for example a WebFinger server on an intranet that
   provides sensitive company information should not allow CORS requests
   from any domain, as that could allow leaking of that sensitive
   information.  WebFinger servers that wish to restrict access to
   information from external entities SHOULD use a more restrictive
   Access-Control-Allow-Origin header.

7. Access Control

   As with all web resources, access to the /.well-known/webfinger
   resource MAY require authentication.  Further, failure to provide
   required credentials MAY result in the server forbidding access or
   providing a different response than had the client authenticated with
   the server.

   Likewise, a server MAY provide different responses to different
   clients based on other factors, such as whether the client is inside
   or outside a corporate network.  As a concrete example, a query
   performed on the internal corporate network might return link
   relations to employee pictures, whereas link relations for employee
   pictures might not be provided to external entities.

Jones, et al.            Expires June 2, 2013                  [Page 15]

Internet-Draft                WebFinger                    December 2012

   Further, link relations provided in a WebFinger server response MAY
   point to web resources that impose access restrictions.  For example,
   the aforementioned corporate server may provide both internal and
   external entities with URIs to employee pictures, but further
   authentication might be required in order for the client to access
   the picture resources if the request comes from outside the corporate

   The decisions made with respect to what set of link relations a
   WebFinger server provides to one client versus another and what
   resources require further authentication, as well as the specific
   authentication mechanisms employed, are outside the scope of this

8. Hosted WebFinger Services

   As with most services provided on the Internet, it is possible for a
   domain owner to utilize "hosted" WebFinger services.  By way of
   example, a domain owner might control most aspects of their domain,
   but use a third-party hosting service for email.  In the case of
   email, mail servers for a domain are identified by MX records.  An MX
   record points to the mail server to which mail for the domain should
   be delivered.  It does not matter to the sending mail server whether
   those MX records point to a server in the destination domain or a
   different domain.

   Likewise, a domain owner might utilize the services of a third party
   to provide WebFinger services on behalf of its users.  Just as a
   domain owner was required to insert MX records into DNS to allow for
   hosted email serves, the domain owner is required to redirect HTTP(S)
   queries to its domain to allow for hosted WebFinger services.

   When a query is issued to /.well-known/webfinger, the web server MUST
   return a 301, 302, or 307 response status code that includes a
   Location header pointing to the location of the hosted WebFinger
   service URL.  The WebFinger service URL does not need to point to
   /.well-known/* on the hosting service provider server.  WebFinger
   clients MUST follow all 301, 302, or 307 redirection requests.

   As an example, assume that example.com's WebFinger services are
   hosted by example.net.  Suppose a client issues a query for
   acct:alice@example.com like this:

     GET /.well-known/webfinger?
                   resource=acct%3Aalice%40example.com HTTP/1.1
     Host: example.com

   The server might respond with this:

Jones, et al.            Expires June 2, 2013                  [Page 16]

Internet-Draft                WebFinger                    December 2012

     HTTP/1.1 307 Temporary Redirect
     Location: http://wf.example.net/example.com/webfinger?
                       resource=acct%3Aalice%40example.com HTTP/1.1

   The client MUST follow the redirection, re-issuing the request to the
   URL provided in the Location header.

9. Security Considerations

   Since this specification utilizes Cross-Origin Resource Sharing
   (CORS) [10], all of the security considerations applicable CORS are
   also applicable to this specification.

   The recommended use of HTTPS is to ensure that information is not
   modified during transit.  It should be appreciated that in
   environments where an HTTPS server is normally available, there
   exists the possibility that a compromised network might have its
   WebFinger server operating on HTTPS replaced with one operating only
   over HTTP.  As such, clients that need to ensure data is not
   compromised SHOULD NOT issue queries over a non-secure connection.
   While Section 5.1 allows for clients that fail to establish an HTTPS
   connection to attempt a query using HTTP, a client and any underlying
   client libraries are not required to re-issue queries using HTTP and
   SHOULD NOT when security for a given application that uses WebFinger
   is paramount.

   When using HTTPS, clients MUST verify that the certificate used on an
   HTTPS connection is valid.

   Service providers and users should be aware that placing information
   on the Internet accessible through WebFinger means that any user can
   access that information.  While WebFinger can be an extremely useful
   tool for allowing quick and easy access to one's avatar, blog, or
   other personal information, users should understand the risks, too.
   If one does not wish to share certain information with the world, do
   not allow that information to be freely accessible through WebFinger
   and do not use any service supporting WebFinger.  Further, WebFinger
   servers MUST NOT be used to provide any personal information to any
   party unless explicitly or implicitly authorized by the person whose
   information is being shared. Implicit authorization can be determined
   by the user's voluntary utilization of a service as defined by that
   service's relevant terms of use or published privacy policy.

   The aforementioned word of caution is perhaps worth emphasizing again
   with respect to dynamic information one might wish to share, such as
   the current location of a user.  WebFinger can be a powerful tool
   used to assemble information about a person all in one place, but
   service providers and users should be mindful of the nature of that
   information shared and the fact that it might be available for the

Jones, et al.            Expires June 2, 2013                  [Page 17]

Internet-Draft                WebFinger                    December 2012

   entire world to see.  Sharing location information, for example,
   would potentially put a person in danger from any individual who
   might seek to inflict harm on that person.

   The easy access to user information via WebFinger was a design goal
   of the protocol, not a limitation.  If one wishes to limit access to
   information available via WebFinger, such as a WebFinger server for
   use inside a corporate network, the network administrator must take
   measures necessary to limit access from outside the network.  Using
   standard methods for securing web resources, network administrators
   do have the ability to control access to resources that might return
   sensitive information.  Further, WebFinger servers can be employed in
   such a way as to require authentication and prevent disclosure of
   information to unauthorized entities.

   Finally, a WebFinger server has no means of ensuring that information
   provided by a user is accurate.  Likewise, neither the server nor the
   client can be absolutely guaranteed that information has not been
   manipulated either at the server or along the communication path
   between the client and server.  Use of HTTPS helps to address some
   concerns with manipulation of information along the communication
   path, but it clearly cannot address issues where the server provided
   incorrect information, either due to being provided false information
   or due to malicious behavior on the part of the server administrator.
   As with any information service available on the Internet, users
   should wary of information received from untrusted sources.

10. IANA Considerations

   This specification registers the "webfinger" well-known URI in the
   Well-Known URI Registry as defined by [3].

   URI suffix:  webfinger

   Change controller:  IETF

   Specification document(s):  RFC QQQ

   Related information:  The response from WebFinger server will be a
   JSON Resource Descriptor (JRD) as described in Section 5.2 of RFC

   [RFC EDITOR: Please replace "QQQ" references in this section with the
   number for this RFC.]

Jones, et al.            Expires June 2, 2013                  [Page 18]

Internet-Draft                WebFinger                    December 2012

11. Acknowledgments

   The authors would like to acknowledge Eran Hammer-Lahav, Blaine Cook,
   Brad Fitzpatrick, Laurent-Walter Goix, Joe Clarke, Michael B. Jones,
   Peter Saint-Andre, Dick Hardt, Tim Bray, and Joe Gregorio for their
   invaluable input.

12. References

12.1. Normative References

   [1]   Bradner, S., "Key words for use in RFCs to Indicate Requirement
         Levels", BCP 14, RFC 2119, March 1997.

   [2]   Fielding, R., Gettys, J., Mogul, J., Frystyk, H.,
         Masinter, L., Leach, P., and T. Berners-Lee, "Hypertext
         Transfer Protocol -- HTTP/1.1", RFC 2616, June 1999.

   [3]   Nottingham, M., Hammer-Lahav, E., "Defining Well-Known Uniform
         Resource Identifiers (URIs)", RFC 5785, April 2010.

   [4]   Nottingham, M., "Web Linking", RFC 5988, October 2010.

   [5]   Crockford, D., "The application/json Media Type for
         JavaScript Object Notation (JSON)", RFC 4627, July 2006.

   [6]   Berners-Lee, T., Fielding, R., and Masinter, L., "Uniform
         Resource Identifier (URI): Generic Syntax", STD 66, RFC 3986,
         January 2005.

   [7]   Duerst, M., "Internationalized Resource Identifiers (IRIs)",
         RFC 3987, January 2005.

   [8]   Saint-Andre, P., "The 'acct' URI Scheme", draft-ietf-appsawg-
         acct-uri-01, October 2012.

   [9]   Duerst, M., Masinter, L., and J. Zawinski, "The 'mailto' URI
         Scheme", RFC 6068, October 2010.

   [10]  Van Kesteren, A., "Cross-Origin Resource Sharing", W3C CORS
         http://www.w3.org/TR/cors/, July 2010.

   [11]  IANA, "Link Relations", http://www.iana.org/assignments/link-

   [12]  IANA, "MIME Media Types",

Jones, et al.            Expires June 2, 2013                  [Page 19]

Internet-Draft                WebFinger                    December 2012

   [13]  Freed, N., Klensin, J., "Media Type Specifications and
         Registration Procedures", RFC 4288, December 2005.

   [14]  Phillips, A., Davis, M., "Tags for Identifying Languages", RFC
         5646, January 2001.

12.2. Informative References

   [15]  Perreault, S., "vCard Format Specification", RFC 6350, August

   [16]  "Transport Independent, Printer/System Interface", IEEE Std
         1284.1-1997, 1997.

   [17]  Sakimura, N., Bradley, J., Jones, M., de Medeiros, B.,
         Mortimore, C., and E. Jay, "OpenID Connect Messages 1.0", June
         2012, http://openid.net/specs/openid-connect-messages-1_0.html.

Author's Addresses

   Paul E. Jones
   Cisco Systems, Inc.
   7025 Kit Creek Rd.
   Research Triangle Park, NC 27709

   Phone: +1 919 476 2048
   Email: paulej@packetizer.com
   IM: xmpp:paulej@packetizer.com

   Gonzalo Salgueiro
   Cisco Systems, Inc.
   7025 Kit Creek Rd.
   Research Triangle Park, NC 27709

   Phone: +1 919 392 3266
   Email: gsalguei@cisco.com
   IM: xmpp:gsalguei@cisco.com

   Joseph Smarr

   Email: jsmarr@google.com

Jones, et al.            Expires June 2, 2013                  [Page 20]