A library for interacting with Control Room work items.

Work items are used for managing data that go through multiple steps and tasks inside a process. Each step of a process receives input work items from the previous step, and creates output work items for the next step.

Item structure

A work item's data payload is JSON and allows storing anything that is serializable. This library by default interacts with payloads that are a dictionary of key-value pairs, which it treats as individual variables. These variables can be exposed to the Robot Framework task to be used directly.

In addition to the data section, a work item can also contain files, which are stored by default in Robocorp Control Room. Adding and using files with work items requires no additional setup from the user.

Loading inputs

The library automatically loads the first input work item, if the library input argument autoload is truthy (default).

After an input has been loaded its payload and files can be accessed through corresponding keywords, and optionally these values can be modified.

E-mail triggering

Since a process can be started in Control Room by sending an e-mail, a body in Text/JSON/YAML/HTML format can be sent as well and this gets attached to the input work item with the rawEmail payload variable. This library automatically parses the content of it and saves into parsedEmail the dictionary transformation of the original e-mail.

If "Parse email" Control Room configuration option is enabled (recommended), then your e-mail is automatically parsed in the work item under the email payload variable, which is a dictionary containing a body holding the final parsed form of the interpreted e-mail body. The payload variable parsedEmail is still available for backwards compatibility reasons and holds the very same body inside the parsedEmail[Body].

E-mail attachments will be added into the work item as files. Read more on: https://robocorp.com/docs/control-room/attended-or-unattended/email-trigger


After starting the process by sending an e-mail with a body like:

{ "message": "Hello world!" }

The robot can use the parsed e-mail body's dictionary:

*** Tasks *** Using Parsed Emails ${mail} = Get Work Item Variable email Set Work Item Variables &{mail}[body] ${message} = Get Work Item Variable message Log ${message} # will print "Hello world!"

The behaviour can be disabled by loading the library with auto_parse_email=${None} or altered by providing to it a dictionary with one "key: value" where the key is usually "email.text" (deprecated "rawEmail", the variable set by Control Room, which acts as source for the parsed (deprecated raw) e-mail data) and the value can be "email.body" (deprecated "parsedEmail", where the parsed e-mail data gets stored into), value which can be customized and retrieved with Get Work Item Variable.

Creating outputs

It's possible to create multiple new work items as an output from a task. With the keyword Create Output Work Item a new empty item is created as a child for the currently loaded input.

All created output items are sent into the input queue of the next step in the process.

Active work item

Keywords that read or write from a work item always operate on the currently active work item. Usually that is the input item that has been automatically loaded when the execution started, but the currently active item is changed whenever the keywords Create Output Work Item or Get Input Work Item are called. It's also possible to change the active item manually with the keyword Set current work item.

Saving changes

While a work item is loaded automatically when a suite starts, changes are not automatically reflected back to the source. The work item will be modified locally and then saved when the keyword Save Work Item is called. This also applies to created output work items.

It is recommended to defer saves until all changes have been made to prevent leaving work items in a half-modified state in case of failures.

Local Development

While Control Room is the default implementation, it can also be replaced with a custom adapter. The selection is based on either the default_adapter argument for the library, or the RPA_WORKITEMS_ADAPTER environment variable. The library has a built-in alternative adapter called FileAdapter for storing work items to disk.

The FileAdapter uses a local JSON file for input work items. It's a list of work items, each of which has a data payload and files.

An example of a local file with one work item:

[ { "payload": { "variable1": "a-string-value", "variable2": ["a", "list", "value"] }, "files": { "file1": "path/to/file.ext" } } ]

Output work items (if any) are saved to an adjacent file with the same name, but with the extension .output.json. You can specify through the "RPA_OUTPUT_WORKITEM_PATH" env var a different path and name for this file.

Simulating the Cloud with Robocorp Code VSCode Extension

If you are developing in VSCode with the Robocorp Code extension, you can utilize the built in local development features described in the Developing with work items locally section of the Using work items development guide.


Robot Framework

In the following example a task creates an output work item, and attaches some variables to it.

*** Settings *** Library RPA.Robocorp.WorkItems *** Tasks *** Save variables to Control Room Create Output Work Item Set work item variables user=Dude mail=address@company.com Save Work Item

In the next step of the process inside a different robot, we can use previously saved work item variables. Also note how the input work item is loaded implicitly when the suite starts.

*** Settings *** Library RPA.Robocorp.WorkItems *** Tasks *** Use variables from Control Room Set task variables from work item Log Variables are now available: s${user}, ${mail}


The library can also be used through Python, but it does not implicitly load the first work item.

import logging from RPA.Robocorp.WorkItems import WorkItems def list_variables(item_id): library = WorkItems() library.get_input_work_item() variables = library.get_work_item_variables() for variable, value in variables.items(): logging.info("%s = %s", variable, value)


Initialize self. See help(type(self)) for accurate signature.