HTTP automations are used when the remote system is an API instead of a file server or inbox.
They support standard API requests, OAuth, pagination, pointer-based fetching, and trigger chains for multi-step workflows.
Basic HTTP Connection
A typical HTTP connection includes an address, method, headers, and optionally a payload:
"connection": {
"type": "http",
"address": "https://api.vendor.example/products",
"method": "GET",
"headers": {
"Authorization": "Bearer {{api_token}}"
}
}
OAuth
For APIs that require OAuth, place an oauth object inside connection. The engine can request tokens, persist them, and refresh them when they expire.
Redirect-based OAuth is also supported for integrations that require a browser authorization step after installation.
Pagination
Pagination is configured inside connection.pagination. Depending on the API, you can paginate by next-token or by page number.
Pointers
Pointers are useful when the API does not expose normal pagination but does let you query a moving range, ID, or timestamp. The automation stores the current pointer value in metadata and reuses it on later runs.
This is commonly used for APIs that support start/end IDs, record offsets, or incremental windows.
Triggers
Triggers are used when you must make one or more preliminary API calls before the final request can succeed.
Examples include:
- Requesting a report job ID, then polling for its completion
- Creating an export request, then downloading the finished file using the returned ID
- Chaining multiple dependent IDs across several API calls
Trigger responses save values such as {{trigger_value}}, {{trigger_value1}}, and {{trigger_value2}} back into the automation flow. The trigger steps run in order, top to bottom.
Validation Note
If you configure triggers, your automation must actually use the trigger placeholder somewhere in the configuration. Otherwise validation will warn that the trigger value is never consumed.
When To Use What
- Use OAuth when the vendor requires token-based auth
- Use pagination when the API returns pages or next tokens
- Use pointers when you need to continue from the last seen ID or range
- Use triggers when the API is multi-step and returns IDs you must reuse later in the flow