​
Introduction

For most applications, the recommended way to use Hanko is by integrating Hanko Elements. In case you want to build a custom frontend instead of using Hanko Elements, for example when building a native mobile app or if Elements do not match your requirements, the Hanko Flow API and the hanko-frontend-sdk are the tools you need for that. The Flow API simplifies frontend development by moving authentication flow logic to the backend. Traditional authentication implementations require frontends to manage complex state transitions, predict next steps, and handle various configuration options. This complexity increases maintenance overhead and makes configuration changes difficult. The Flow API solves this by maintaining all authentication state on the backend. Your frontend only needs to render the current state and execute available actions, without tracking user inputs or predicting what comes next. Key takeaways include:
  1. State-Driven Flow: Each API response represents a state, which determines the available actions and required user inputs.
  2. Dynamic Input Validation: Inputs come with metadata for client-side validation, while the backend ensures robust validation.
  3. Error Handling: Detailed error annotations in responses guide the frontend in displaying actionable feedback to users.

​
Understanding states and actions

Every Flow API response contains all information needed to render the appropriate UI. Each response represents a specific state in the authentication flow, and each state defines available actions that users can perform. Your frontend implementation requires a handler function for each possible state. When you receive a new response from the backend, the corresponding handler processes it and renders the UI based on the available actions. Each action includes:
  • inputs: Defines what data to collect from the user
  • validation rules: Specifies requirements for input values
  • execution URL: The endpoint to send the action request to
To execute an action, send a POST request to the action’s URL with the collected input data. The backend processes the action and returns the next state. Some responses include a payload with additional data specific to that state, such as user information or configuration details needed for UI rendering. The diagram shows the structure of a flow response:

​
Initializing flows

The Flow API supports three main flows, each initialized with an empty POST request:
  • /login - User authentication flow
  • /registration - User registration flow
  • /profile - User profile management flow
Refer to the Flow API documentation for complete details on available states and actions in each flow. This guide uses an example /example-flow endpoint to demonstrate the concepts. Here’s how to initialize and work with any flow:
1

Initialize a new flow

Send an empty POST request to the the flow URL, /example-flow in this example.
2

Obtain the response body

When the API successfully created a new flow, the response will look like this:
{
  "name": "example-state",
  "status": 200,
  "csrf_token": "a-random-token",
  "actions": {
    "example-action": {
      "action": "example-action",
      "description": "This action is an example.",
      "href": "/example-flow?action=example-action@example-flow-id",
      "inputs": {
        "example-input-field": {
          "name": "example-input-field",
          "type": "string",
          "minLength": 16,
          "maxLength": 32,
          "required": true,
          "hidden": false
        }
      }
    }
  }
}
The response object represents a state called ‘example-state’. It includes a single action, ‘example-action’.

​
Interacting with flows

After receiving a flow response, you process the current state by calling the appropriate handler function. This handler renders the UI based on the state and available actions. 
  const state = await fetch(...)

  switch (state.name) {
    case "example-state":
      // Render a page that represents the 'example-state', containing a
      // string input field, where the user can enter a value for the
      // 'example-input-field' (see previous section).
      break;
    // Add more cases for different states.
    default:
      console.log(`Missing handler for state: ${state.name}`);
  }
To execute an action, collect the input values from your form and create a request body with the structure shown in the diagram:
According to the example reponse object, we would proceed as follows to execute the ‘example-action’:
1

Generate the request object

Let’s assume that we’ve obtained an input value ‘example input value’ for the input field ‘example-input-field’. The request body to be generated would look like this:
{
    "input_data": {
        "example-input-field": "example input value"
    },
    "csrf_token": "a-random-token"
}
Note that the csrf_token from the previous state must be included in each request.
2

Execute the action

Send the object via a POST request to the href provided by the action. In this case, the href is /example-flow?action=example-action@example-flow-id.
3

Obtain the response body

When the request was successful, you receive an object in the same manner, representing the next state:
{
  "name": "next-state",
  "status": 200,
  "csrf_token": "a-new-random-token",
  "actions": { ... } // a new set of available actions
}

​
Handling validation errors

The Flow API includes detailed validation metadata for each input field, such as minimum and maximum length requirements (min_length and max_length). While you can implement client-side validation using this metadata, the backend always performs final validation and returns detailed error information when validation fails:
1

Send invalid input values

Let’s assume that we’ve generated an invalid value for ‘example-input-field’:
{
    "input_data": {
        "example-input-field": "too short" // less than 16 characters
    },
    "csrf_token": "a-random-token"
}
2

Execute the action

Send the object via a POST request to the href provided by the action. Again, the href would be /example-flow?action=example-action@example-flow-id.
3

Obtain the response body

The API would return an object representing the same state, ‘example-state’, but with error annotations indicating the validation issue:
{
  "name": "example-state",
  "status": 400,
  "csrf_token": "a-random-token",
  "error": {
    "code": "form_data_invalid_error",
    "message": "Form data invalid."
  },
  "actions": {
    "example-action": {
      "action": "example-action",
      "description": "This action is an example.",
      "href": "/example-flow?action=example-action@example-flow-id",
      "inputs": {
        "example-input-field": {
          "name": "example-input-field",
          "type": "string",
          "minLength": 16,
          "maxLength": 32,
          "required": true,
          "hidden": false,
          "error": {
            "code": "value_too_short_error",
            "message": "The value is too short."
          }
        }
      }
    }
  }
}
As you can see, the status field has changed to ‘400’ (Bad Request), indicating that the request was invalid. A global error object has been added with the code ‘form_data_invalid_error’ and the message ‘Form data invalid’. Additionally, the invalid input field now includes an error object with the code ‘value_too_short_error’, and the message ‘The value is too short’.

​
Handling technical errors

Technical errors (non-4xx HTTP errors) result in an error state that indicates the flow cannot continue. These typically occur due to server issues, network problems, or system failures. Here’s how such responses appear: 
{
    "name": "error",
    "status": 500,
    "error": {
        "code": "technical_error",
        "message": "Something went wrong."
    }
}
When you encounter an error state, display the error message to the user and provide an option to restart the authentication process by initializing a new flow.

​
Configuring flows

Flow behavior is highly configurable to match your application’s requirements: Your configuration determines which states and actions are available in each flow. Different configurations can enable or disable specific authentication methods, require additional verification steps, or change the overall flow structure. When building custom UIs, you have two approaches:
  1. Full support: Implement handlers for all possible states and actions to support every available feature
  2. Selective support: Configure your Hanko project to disable unused features and implement only the handlers you need
After making configuration changes, always test your application thoroughly and add any missing state handlers to ensure proper functionality.

​
Conclusion and outlook

This guide has shown you the core concepts of the Flow API through practical examples:
  • Initialize flows with empty POST requests to flow endpoints
  • Interpret responses by understanding states, actions, and input requirements
  • Execute actions by collecting user input and sending proper request payloads
  • Handle validation errors with detailed field-level error information
  • Manage technical errors by implementing proper error recovery flows
For complete details on available states and actions in each flow, consult the Flow API documentation. The Flow API continues to evolve with additional features and improved tooling to make authentication integration even simpler.