RT::Extension::REST2 - Adds a modern REST API to RT under /REST/2.0/


perl Makefile.PL
make install

May need root permissions

Edit your /opt/rt4/etc/

Add this line:

Clear your mason cache
    rm -rf /opt/rt4/var/mason_data/obj
Restart your webserver



To make it easier to authenticate to REST2, we recommend installing RT::Authen::Token. Visit "Logged in as ___" -> Settings -> Auth Tokens. Create an Auth Token, give it any description (such as "REST2 with curl"). Make note of the authentication token it provides to you.

For other authentication options see the section "Authentication Methods" below.


Run the following in a terminal, filling in XX_TOKEN_XX from the auth token above and XX_RT_URL_XX with the URL for your RT instance.

    curl -H 'Authorization: token XX_TOKEN_XX' 'XX_RT_URL_XX/REST/2.0/queues/all'

This does an authenticated request (using the Authorization HTTP header with type token) for all of the queues you can see. You should see a response, typical of search results, like this:

       "total" : 1,
       "count" : 1,
       "page" : 1,
       "per_page" : 20,
       "items" : [
             "type" : "queue",
             "id" : "1",
             "_url" : "XX_RT_URL_XX/REST/2.0/queue/1"

This format is JSON, which is a format for which many programming languages provide libraries for parsing and generating.

(If you instead see a response like {"message":"Unauthorized"} that indicates RT couldn't process your authentication token successfully; make sure the word "token" appears between "Authorization:" and the auth token that RT provided to you)

You can request one of the provided _urls to get more information about that queue.

    curl -H 'Authorization: token XX_TOKEN_XX' 'XX_QUEUE_URL_XX'

This will give a lot of information, like so:

       "id" : 1,
       "Name" : "General",
       "Description" : "The default queue",
       "Lifecycle" : "default",
       "CustomFields" : {},
       "_hyperlinks" : [
             "id" : "1",
             "ref" : "self",
             "type" : "queue",
             "_url" : "XX_RT_URL_XX/REST/2.0/queue/1"
             "ref" : "history",
             "_url" : "XX_RT_URL_XX/REST/2.0/queue/1/history"
             "ref" : "create",
             "type" : "ticket",
             "_url" : "XX_RT_URL_XX/REST/2.0/ticket?Queue=1"

Of particular note is the _hyperlinks key, which gives you a list of related resources to examine (following the principle). For example an entry with a ref of history lets you examine the transaction log for a record. You can implement your REST API client knowing that any other hypermedia link with a ref of history has the same meaning, regardless of whether it's the history of a queue, ticket, asset, etc.

Another ref you'll see in _hyperlinks is create, with a type of ticket. This of course gives you the URL to create tickets in this queue. Importantly, if your user does not have the CreateTicket permission in this queue, then REST2 would simply not include this hyperlink in its response to your request. This allows you to dynamically adapt your client's behavior to its presence or absence, just like the web version of RT does.

Creating Tickets

Let's use the _url from the create hyperlink with type ticket.

To create a ticket is a bit more involved, since it requires providing a different HTTP verb (POST instead of GET), a Content-Type header (to tell REST2 that your content is JSON instead of, say, XML), and the fields for your new ticket such as Subject. Here is the curl invocation, wrapped to multiple lines for readability.

    curl -X POST
         -H "Content-Type: application/json"
         -d '{ "Subject": "hello world" }'
         -H 'Authorization: token XX_TOKEN_XX'

If successful, that will provide output like so:

        "_url" : "XX_RT_URL_XX/REST/2.0/ticket/20",
        "type" : "ticket",
        "id"   : "20"

(REST2 also produces the status code of 201 Created with a Location header of the new ticket, which you may choose to use instead of the JSON response)

We can fetch that _url to continue working with this newly-created ticket. Request the ticket like so (make sure to include the -i flag to see response's HTTP headers).

    curl -i -H 'Authorization: token XX_TOKEN_XX' 'XX_TICKET_URL_XX'

You'll first see that there are many hyperlinks for tickets, including one for each Lifecycle action you can perform, history, comment, correspond, etc. Again these adapt to whether you have the appropriate permissions to do these actions.

Additionally you'll see an ETag header for this record, which can be used for conflict avoidance ( We'll first try updating this ticket with an invalid ETag to see what happens.

Updating Tickets

For updating tickets we use the PUT verb, but otherwise it looks much like a ticket creation.

    curl -X PUT
         -H "Content-Type: application/json"
         -H "If-Match: invalid-etag"
         -d '{ "Subject": "trial update" }'
         -H 'Authorization: token XX_TOKEN_XX'

You'll get an error response like {"message":"Precondition Failed"} and a status code of 412. If you examine the ticket, you'll also see that its Subject was not changed. This is because the If-Match header advises the server to make changes if and only if the ticket's ETag matches what you provide. Since it differed, the server refused the request and made no changes.

Now, try the same request by replacing the value "invalid-etag" in the If-Match request header with the real ETag you'd received when you requested the ticket previously. You'll then get a JSON response like:

    ["Ticket 1: Subject changed from 'hello world' to 'trial update'"]

which is a list of messages meant for displaying to an end-user.

If you GET the ticket again, you'll observe that the ETag header now has a different value, indicating that the ticket itself has changed. This means if you were to retry the PUT update with the previous (at the time, expected) ETag you would instead be rejected by the server with Precondition Failed.

You can use ETag and If-Match headers to avoid race conditions such as two people updating a ticket at the same time. Depending on the sophistication of your client, you may be able to automatically retry the change by incorporating the changes made on the server (for example adding time worked can be automatically be recalculated).

You may of course choose to ignore the ETag header and not provide If-Match in your requests; RT doesn't require its use.


RT's REST2 API provides the tools you need to build robust and dynamic integrations. Tools like ETag/If-Match allow you to avoid conflicts such as two people taking a ticket at the same time. Using JSON for all data interchange avoids problems caused by parsing text. Hypermedia links inform your client application of what the user has the ability to do.

Careful readers will see that, other than our initial entry into the system, we did not generate any URLs. We only followed links, just like you do when browsing a website on your computer. We've better decoupled the client's implementation from the server's REST API. Additionally, this system lets you be informed of new capabilities in the form of additional hyperlinks.

Using these tools and principles, REST2 will help you build rich, robust, and powerful integrations with the other applications and services that your team uses.


Currently provided endpoints under /REST/2.0/ are described below. Wherever possible please consider using _hyperlinks hypermedia controls available in response bodies rather than hardcoding URLs.

For simplicity, the examples below omit the extra options to curl for SSL like --cacert.


    GET /tickets?query=<TicketSQL>
        search for tickets using TicketSQL

    GET /tickets?simple=1;query=<simple search query>
        search for tickets using simple search syntax

    POST /tickets
        search for tickets with the 'query' and optional 'simple' parameters

    POST /ticket
        create a ticket; provide JSON content

    GET /ticket/:id
        retrieve a ticket

    PUT /ticket/:id
        update a ticket's metadata; provide JSON content

    DELETE /ticket/:id
        set status to deleted

    POST /ticket/:id/correspond
    POST /ticket/:id/comment
        add a reply or comment to the ticket

    GET /ticket/:id/history
        retrieve list of transactions for ticket

    POST /tickets/bulk
        create multiple tickets; provide JSON content(array of hashes)

    PUT /tickets/bulk
        update multiple tickets' metadata; provide JSON content(array of hashes)

Ticket Examples

Below are some examples using the endpoints above.

    # Create a ticket, setting some custom fields
    curl -X POST -H "Content-Type: application/json" -u 'root:password'
        -d '{ "Queue": "General", "Subject": "Create ticket test",
            "Requestor": "", "Cc": "",
            "Content": "Testing a create",
            "CustomFields": {"Severity": "Low"}}'

    # Update a ticket, with a custom field update
    curl -X PUT -H "Content-Type: application/json" -u 'root:password'
        -d '{ "Subject": "Update test", "CustomFields": {"Severity": "High"}}'

    # Correspond a ticket
    curl -X POST -H "Content-Type: application/json" -u 'root:password'
        -d '{ "Content": "Testing a correspondence", "ContentType": "text/plain" }'

    # Comment a ticket
    curl -X POST -H "Content-Type: text/plain" -u 'root:password'
        -d 'Testing a comment'

    # Create an Asset
    curl -X POST -H "Content-Type: application/json" -u 'root:password'
        -d '{"Name" : "Asset From Rest", "Catalog" : "General assets", "Content" : "Some content"}'

    # Search Assets
    curl -X POST -H "Content-Type: application/json" -u 'root:password'
    -d '[{ "field" : "id", "operator" : ">=", "value" : 0 }]'


    GET /transactions?query=<JSON>
    POST /transactions
        search for transactions using L</JSON searches> syntax

    GET /ticket/:id/history
    GET /queue/:id/history
    GET /queue/:name/history
    GET /asset/:id/history
    GET /user/:id/history
    GET /user/:name/history
    GET /group/:id/history
        get transactions for record

    GET /transaction/:id
        retrieve a transaction

Attachments and Messages

    GET /attachments?query=<JSON>
    POST /attachments
        search for attachments using L</JSON searches> syntax

    GET /transaction/:id/attachments
        get attachments for transaction

    GET /attachment/:id
        retrieve an attachment


    GET /queues/all
        retrieve list of all queues you can see

    GET /queues?query=<JSON>
    POST /queues
        search for queues using L</JSON searches> syntax

    POST /queue
        create a queue; provide JSON content

    GET /queue/:id
    GET /queue/:name
        retrieve a queue by numeric id or name

    PUT /queue/:id
    PUT /queue/:name
        update a queue's metadata; provide JSON content

    DELETE /queue/:id
    DELETE /queue/:name
        disable queue

    GET /queue/:id/history
    GET /queue/:name/history
        retrieve list of transactions for queue


    GET /assets?query=<JSON>
    POST /assets
        search for assets using L</JSON searches> syntax

    POST /asset
        create an asset; provide JSON content

    GET /asset/:id
        retrieve an asset

    PUT /asset/:id
        update an asset's metadata; provide JSON content

    DELETE /asset/:id
        set status to deleted

    GET /asset/:id/history
        retrieve list of transactions for asset


    GET /catalogs/all
        retrieve list of all catalogs you can see

    GET /catalogs?query=<JSON>
    POST /catalogs
        search for catalogs using L</JSON searches> syntax

    POST /catalog
        create a catalog; provide JSON content

    GET /catalog/:id
    GET /catalog/:name
        retrieve a catalog by numeric id or name

    PUT /catalog/:id
    PUT /catalog/:name
        update a catalog's metadata; provide JSON content

    DELETE /catalog/:id
    DELETE /catalog/:name
        disable catalog


    GET /users?query=<JSON>
    POST /users
        search for users using L</JSON searches> syntax

    POST /user
        create a user; provide JSON content

    GET /user/:id
    GET /user/:name
        retrieve a user by numeric id or username (including its memberships and whether it is disabled)

    PUT /user/:id
    PUT /user/:name
        update a user's metadata (including its Disabled status); provide JSON content

    DELETE /user/:id
    DELETE /user/:name
        disable user

    GET /user/:id/history
    GET /user/:name/history
        retrieve list of transactions for user


    GET /groups?query=<JSON>
    POST /groups
        search for groups using L</JSON searches> syntax

    POST /group
        create a (user defined) group; provide JSON content

    GET /group/:id
        retrieve a group (including its members and whether it is disabled)

    PUT /group/:id
        update a groups's metadata (including its Disabled status); provide JSON content

    DELETE /group/:id
        disable group

    GET /group/:id/history
        retrieve list of transactions for group

User Memberships

    GET /user/:id/groups
    GET /user/:name/groups
        retrieve list of groups which a user is a member of

    PUT /user/:id/groups
    PUT /user/:name/groups
        add a user to groups; provide a JSON array of groups ids

    DELETE /user/:id/group/:id
    DELETE /user/:name/group/:id
        remove a user from a group

    DELETE /user/:id/groups
    DELETE /user/:name/groups
        remove a user from all groups

Group Members

    GET /group/:id/members
        retrieve list of direct members of a group

    GET /group/:id/members?recursively=1
        retrieve list of direct and recursive members of a group

    GET /group/:id/members?users=0
        retrieve list of direct group members of a group

    GET /group/:id/members?users=0&recursively=1
        retrieve list of direct and recursive group members of a group

    GET /group/:id/members?groups=0
        retrieve list of direct user members of a group

    GET /group/:id/members?groups=0&recursively=1
        retrieve list of direct and recursive user members of a group

    PUT /group/:id/members
        add members to a group; provide a JSON array of principal ids

    DELETE /group/:id/member/:id
        remove a member from a group

    DELETE /group/:id/members
        remove all members from a group

Custom Fields

    GET /customfields?query=<JSON>
    POST /customfields
        search for custom fields using L</JSON searches> syntax

    GET /customfield/:id
        retrieve a custom field

Custom Roles

    GET /customroles?query=<JSON>
    POST /customroles
        search for custom roles using L</JSON searches> syntax

    GET /customrole/:id
        retrieve a custom role


    GET /
        produces this documentation

    GET /rt
        produces system information

JSON searches

Some resources accept a basic JSON structure as the search conditions which specifies one or more fields to limit on (using specified operators and values). An example:

    curl -si -u user:pass -XPOST --data-binary '
            { "field":    "Name",
              "operator": "LIKE",
              "value":    "Engineering" },

            { "field":    "Lifecycle",
              "value":    "helpdesk" }

The JSON payload must be an array of hashes with the keys field and value and optionally operator.

Results can be sorted by using multiple query parameter arguments orderby and order. Each orderby query parameter specify a field to be used for sorting results. If the request includes more than one orderby query parameter, results are sorted according to corresponding fields in the same order than they are specified. For instance, if you want to sort results according to creation date and then by id (in case of some items have the same creation date), your request should specify ?orderby=Created&orderby=id. By default, results are sorted in ascending order. To sort results in descending order, you should use order=DESC query parameter. Any other value for order query parameter will be treated as order=ASC, for ascending order. The order of the order query parameters should be the same as the orderby query parameters. Therefore, if you specify two fields to sort the results (with two orderby parameters) and you want to sort the second field by descending order, you should also explicitely specify order=ASC for the first field: orderby=Created&order=ASC&orderby=id&order=DESC. orderby and order query parameters are supported in both JSON and TicketSQL searches.

Results are returned in the format described below.

Example of plural resources (collections)

Resources which represent a collection of other resources use the following standard JSON format:

       "count" : 20,
       "page" : 1,
       "per_page" : 20,
       "total" : 3810,
       "items" : [
          { … },
          { … },

Each item is nearly the same representation used when an individual resource is requested.


All plural resources (such as /tickets) require pagination, controlled by the query parameters page and per_page. The default page size is 20 items, but it may be increased up to 100 (or decreased if desired). Page numbers start at 1.


When fetching search results you can include additional fields by adding a query parameter fields which is a comma seperated list of fields to include. You must use the camel case version of the name as included in the results for the actual item.

You can use additional fields parameters to expand child blocks, for example (line wrapping inserted for readability):


Says that in the result set for tickets, the extra fields for Owner, Status, Created, Subject and Queue should be included. But in addition, for the Queue block, also include Name and Description. The results would be similar to this (only one ticket is displayed):

   "items" : [
         "Subject" : "Sample Ticket",
         "id" : "2",
         "type" : "ticket",
         "Owner" : {
            "id" : "root",
            "_url" : "XX_RT_URL_XX/REST/2.0/user/root",
            "type" : "user"
         "_url" : "XX_RT_URL_XX/REST/2.0/ticket/2",
         "Status" : "resolved",
         "Created" : "2018-06-29:10:25Z",
         "Queue" : {
            "id" : "1",
            "type" : "queue",
            "Name" : "General",
            "Description" : "The default queue",
            "_url" : "XX_RT_URL_XX/REST/2.0/queue/1"
      { … },

If the user performing the query doesn't have rights to view the record (or sub record), then the empty string will be returned.

For single object URLs like /ticket/:id, as it already contains all the fields by default, parameter "fields" is not needed, but you can still use additional fields parameters to expand child blocks:


Authentication Methods

Authentication should always be done over HTTPS/SSL for security. You should only serve up the /REST/2.0/ endpoint over SSL.

Basic Auth

Authentication may use internal RT usernames and passwords, provided via HTTP Basic auth. Most HTTP libraries already have a way of providing basic auth credentials when making requests. Using curl, for example:

    curl -u 'username:password' /path/to/REST/2.0

Token Auth

You may use the RT::Authen::Token extension to authenticate to the REST 2 API. Once you've acquired an authentication token in the web interface, specify the Authorization header with a value of "token" like so:

    curl -H 'Authorization: token …' /path/to/REST/2.0

If the library or application you're using does not support specifying additional HTTP headers, you may also pass the authentication token as a query parameter like so:

    curl /path/to/REST/2.0?token=…

Finally, you may reuse an existing cookie from an ordinary web session to authenticate against REST2. This is primarily intended for interacting with REST2 via JavaScript in the browser. Other REST consumers are advised to use the alternatives above.

Conditional requests (If-Modified-Since, If-Match)

You can take advantage of the Last-Modified headers returned by most single resource endpoints. Add a If-Modified-Since header to your requests for the same resource, using the most recent Last-Modified value seen, and the API may respond with a 304 Not Modified. You can also use HEAD requests to check for updates without receiving the actual content when there is a newer version. You may also add an If-Unmodified-Since header to your updates to tell the server to refuse updates if the record had been changed since you last retrieved it.

ETag, If-Match, and If-None-Match work similarly to Last-Modified, If-Modified-Since, and If-Unmodified-Since, except that they don't use a timestamp, which has its own set of tradeoffs. ETag is an opaque value, so it has no meaning to consumers (unlike timestamps). However, timestamps have the disadvantage of having a resolution of seconds, so two updates happening in the same second would produce incorrect results, whereas ETag does not suffer from that problem.

Status codes

The REST API uses the full range of HTTP status codes, and your client should handle them appropriately.


Best Practical Solutions, LLC <>


All bugs should be reported via email to or via the web at


This software is Copyright (c) 2015-2019 by Best Practical Solutions, LLC.

This is free software, licensed under:

The GNU General Public License, Version 2, June 1991