Skip to main content

Setup backend

Port's self-service actions and automations support a variety of backends that can be used when triggered.

The process contains the following steps:

  1. The action is triggered in Port - the trigger can either be a user executing a self-service action via the UI, or an automation triggering an action.
  2. The payload is sent to your backend - the payload, as defined by the action's creator, is sent to your backend. The backend can be a URL, a dedicated Kafka topic or one of your CI/CD workflows/pipelines.
  3. Your backend receives the payload and handles the request - depending on the action, your backend might open a PR, create a cloud resource, provision a new environment, or perform any other logic you would like.
  4. Your backend updates Port on the status of the execution - You can enrich the action run object in Port by adding logs, attaching links to other workflows or pipelines that help fullfil the request and add a final success/fail status once the action is complete.

Define the backend

UI support

Currently, only self-service actions can be created via a wizard in the UI.
Automations can be created using a JSON definition, which can be found here.

The action's backend is defined under the Backend tab of the action creation form in Port's UI.
Let's break the definition down to two parts:

Define your backend's type and metadata

In this section we provide information about the backend logic and its location, so that Port can access and run it.

Here is an example of a Github workflow backend configuration:


Depending on the backend you choose, you will need to provide different configuration parameters.
See the list of supported backends below for more information.

Define the action's payload

When creating a self-service action or automation, you can construct a JSON payload that will be sent to your backend upon every execution. You can use this to send data about the action that you want your backend to have.

The payload is defined using JSON, and accessing your data is done using jq, wrapping each expression with {{ }}.
For example, this payload contains a timestamp of the execution, and the execution's run id (unique to each execution):

{
"execution_time": "{{ .trigger.at }}",
"port_context": {
"run_id": "{{ .run.id }}"
}
}

Trigger data

When a self-service action or automation is executed, Port creates an actionRun object that contains data about the execution.

This entire object is accessible to you when constructing the payload, and due to the difference in triggers, it is structured a bit differently for self-service actions and automations:

Here is an example of trigger data for a self-service action:

{
"inputs": {
"microservice_name": "string",
"microservice_description": "string",
"language": "string",
"version": "string",
"core": "string",
"features": "string"
},
"trigger": {
"by": {
"orgId": "<Your organization's id>",
"userId": "<Executing user's id>",
"user": {
"email": "<Executing user's email>",
"firstName": "<Executing user's firstName>",
"lastName": "<Executing user's lastName>",
"phoneNumber": "<Executing user's phoneNumber>",
"picture": "",
"providers": [],
"status": "ACTIVE",
"id": "<Executing user's id>",
"createdAt": "2024-06-06T05:21:00.565Z",
"updatedAt": "2024-06-06T05:21:00.565Z"
}
},
"origin": "UI",
"at": "2024-06-06T05:21:00.565Z",
"operation": "CREATE"
},
"event": null,
"entity": {},
"action": {
"identifier": "Microservice_scaffold_a_microservice",
"blueprint": "Microservice",
"encryptedProperties": []
},
"run": {
"id": "<The current run's id>"
}
}

You can access any value in this structure and add it to the payload. For example, to add the executing user's name to the payload, you can use the following expression:

{
"executing_user_email": "{{.trigger.by.user.email}}"
}

Use the Test JQ button in the bottom-left corner to test your expressions against your action and ensure you are sending the correct data.

Pro tip

You can use the jq expression {{ . }} when testing to see the entire available object, and then drill down to the specific data you need.

spreadValue() function

You can use the spreadValue() function to add multiple keys to the root of the payload at once. This function will spread all of the keys under a given object.
A common use case for this function is to add all of the user inputs to the payload:

{
"{{ spreadValue() }}": "{{ .inputs }}"
}

This will add all of the action's user inputs to the root of the payload, so that they can be accessed directly by your backend.

Using jq expressions in keys

The keys in the payload can also be jq expressions.
For example, the following expression will add the ref key to the payload only if a ref input was provided when executing the action:

{
"{{if (.inputs | has(\"ref\")) then \"ref\" else null end}}": "{{.inputs.ref}}"
}
Using jq in keys

Note that if a key in the payload evaluates to null for any reason, the entire expression (key + value) will be ommitted from the payload.

Supported backends

JSON structure

The action's backend is defined under the invocationMethod object:

{
"identifier": "unique_id",
"title": "Title",
"description": "Action description",
"trigger": {
"type": "self-service",
"operation": "CREATE"
"userInputs": {
"properties": {
"property1": {
"type": "string",
"title": "Property title",
"default": "default value"
}
}
},
},
"invocationMethod": {
"type": "WEBHOOK",
"url": "https://example.com"
},
}

Invocation method structure fields

The type field defines the action's backend type, and can have one of the following values: WEBHOOK, GITHUB, GITLAB, KAFKA, UPSERT_ENTITY.

Depending on the backend type you choose, the available fields will be different:

invocationMethod.type should be set to WEBHOOK.

FieldTypeDescriptionExample values
agentbooleanDefines whether to use Port Agent for execution or not.true or false
urlstringDefines the webhook URL to which Port will send the action via an HTTP POST request.https://example.com
headersstringAn object containing the payload headers to be sent to the webhook in each execution, in "key":"value" pairs.
bodystringDefines the payload that will be sent to the backend upon execution of the action.