What Is a Step?

Steps are specific actions within a pipeline. Here are some common actions you can complete with steps:

  • Log in a user into the app
  • Get data from an external data source
  • Send HTTP requests to an external server
  • Alter the structure of a javascript object

See step reference for information on the available step types.

Extension Step

In code, a step is a Node.js module that returns a single function with a specified interface.

Each step function has two important parameters: context and input.


The context is an object that lets you access everything the step needs for execution except for the steps’ input. The context object includes a logger, a storage for several different scopes, settings of the application, general metadata, and the extension config.


Each step function has a context-aware logger available as context.log. The logger is based on the Bunyan logger. You can use it to log the following common log levels:

context.log.trace(additionalDataObject, msg)
context.log.debug(additionalDataObject, msg), msg)
context.log.warn(additionalDataObject, msg)
context.log.error(additionalDataObject, msg)
context.log.fatal(additionalDataObject, msg)

In the production environment, the log level is set to info, meaning messages equal to or higher than info are logged. In the sandbox environment, the log level is set to debug.


Sometimes an extension or the step of an extension needs to store data or retrieve data from storage.

You can use data storage when data must be persistent between multiple pipeline requests. For example, you can store data when a user logs in to a service and receives an access token. You put the token data into storage. Later, pipeline requests with steps that need to access the service can reuse the stored token.

Storage is scoped by extension. When there are steps of two different extensions within a pipeline, those steps can only access the storage of their respective extension.

The context object provides three different storage types: the extension-, device-, and user storage.

Storage name Usage Scoped by Availability
extension when you need to store data in an application-wide scope extension always
device when you need to store data associated with a device extension and device always
user when you need to store data across devices associated with a user extension and user when user is logged in

See storage reference.

Each storage type provides a simple key/value storage and a map storage. You can store any value (strings, numbers, arrays, objects) for a key. You can access complete maps or access items directly from map storage.

Key/value and map storage are not related to each other: the get(‘foo’) method of the key/value store won’t return the same result as the get(‘foo’) method of the map store.


The Meta object provides basic request information, such as the app id of the device where the request originated. The Meta object includes the following properties:

  • appId: your application’s id
  • deviceId: the current device id
  • userId: the current user’s id (if logged in, otherwise null)
  • headers: HTTP request headers if pipeline request originated from browser; undefined if not from browser
  • cookies: HTTP request cookies if pipeline request originated from browser; undefined if not from browser


The config object of the context contains extension-wide config values for the step. These configurations are written by the SDK during the deployment of the application. When developing on a sandbox app with the SDK, these configs are stored in the config.json files that are created when altering the extension-config.json file.

See SDK automations and/or extension config.


The input parameter is an object whose content is defined by the pipeline definition.

See pipeline reference and/or the shared object.


The exported function of a step can return in three ways: call a callback such as:

  • cb(err, output)
  • return a promise, which resolves if everything ran successfully or rejects with an error if something went wrong
  • return the output object directly or throw an error in case you use the async/await way of implementing the function

The return object has to meet the output specified within the pipeline file.

Shared Object

Pipelines and their steps share an object to get input and put output into the object.

The image demonstrates how the shared object works within a pipeline.

The pipeline requires three inputs and returns two outputs. The initial input of the pipeline gets put into the shared object.

While processing the pipeline, steps can read inputs from and write outputs to the shared object. This enables a step to use the output of any former step (not just the direct neighbors) as an input.

Storing and Retrieving Data From the Shared Storage Object

The shared object is a storage where pipelines and steps can put data into or retrieve data. Inputs have names (“key”: “input1”) for the usage within the step, but are stored as a separate id within the shared object.

For example, a pipeline has the input productId. The value of productId will be stored as id 1 in a shared object. A step requiring the data that was stored can now get the data by getting the value of id 1 and then map this value into a new input with the name pId. You can name inputs or outputs anything because they do not affect the naming in the shared object.


Here is the input:
  “input”: [
    {“key”: “productId”, “id”:1}

The shared object contains id 1 which we expect to see in the request parameter productId when the pipeline is called.

Step input:
  “input”: [
    {“key”: “pId”, “id”:1}

The step takes id 1 from the shared input and and maps it to the input key pId.

Step function:
module.exports = async (context, input) => {
  return {p: input.pId}

The step function takes the input key pId and sets its value to the output key p.

Step output:
  “output”: [
    {“key”: “p”, “id”:2}

The value of output key p maps to id 2 in the shared object. The shared object now contains two ids (1 and 2) with the same value.

Pipeline output:
  “output”: [
    {“key”: “productId”, “id”:2}

The pipeline returns what is stored in id 2 in the shared object as the property productId in a result object.


Every step has an input, an output, and a type.

The type can be one of the following: extension, auth, conditional, errorCatchExtension, pipeline, or staticValue.

See steps to learn more about step types.

Both input and output are arrays of objects. These objects define a specific input or output element. This element describes the mapping between the step and the shared object. The element itself consists of two properties: key and id.

Example: {“key”: “productId”, “id”: “1”}


The key value is accessible by the step. For example, an extension step can access the value of key: productId with input.productId inside of the step function.

The id describes where the step gets the input for its key: productId. In the Shared Object product id example, this is the value of id: 1 within the shared object.


The key is what is returned by the step. For example, an extension step can return the value of key: productId with return {productId: “PID1”} inside of the step function.

The id describes where the step puts the value of key: productId. In the Shared Object product id example this is the value of id: 1 within the shared object.

  “type”: “extension”,
  “input”: [
    {“key”: “input1”: “id”:1},
    {“key”: “input2”: “id”:2}
  “output”: [
    {“key”: “output1”: “id”:10},
    {“key”: “output2”: “id”:20}  
  … (other step type dependent properties)

for other step type dependent properties: see steps.