Configure and test Actions

Create an Action

1. Navigate to Console > Actions.
2. Select Post first-factor verification or Post sign-in.
3. Implement the runAction function in the script editor.
4. Under Data source, review the event and result types, configure environment variables, and find an example for fetching external data.
5. Under Test context, adjust the sample event and run the script.
6. Under Settings, enable the Action and choose its script-error behavior.
7. Save the Action.

Only a saved and enabled Action runs in production.

Implement runAction

Keep the entry function name as runAction. It receives a single payload object:

const runAction = async ({ event, environmentVariables = {} }) => {
  return;
};

To decline or continue without a user update, return the no-op value supported by the specific action type.

Fetch external data

Use the injected fetch function to call an external API. For example, a Post sign-in Action can fetch a user profile:

const runAction = async ({ event, environmentVariables = {} }) => {
  const response = await fetch(environmentVariables.PROFILE_API_URL, {
    headers: {
      authorization: `Bearer ${environmentVariables.PROFILE_API_TOKEN}`,
    },
  });

  if (!response.ok) {
    throw new Error(`Profile API returned ${response.status}`);
  }

  const profile = await response.json();

  return {
    action: 'updateUser',
    user: {
      name: profile.name,
    },
  };
};

Actions run in the authentication request path. Keep external services fast and highly available, and assume that a user may retry the sign-in. Logto does not retry an Action or fall back from the Logto Cloud remote runner to local execution.

Use environment variables

Use environment variables for values that should not be hardcoded in the script, such as API URLs, tokens, and feature settings:

const { API_URL, API_TOKEN } = environmentVariables;

Environment variables are part of the Action configuration and are visible to administrators who can read that configuration. Restrict Action-management access and never include secrets in the returned result or an error message.

Supported user patch

An Action can return only these user fields:

Field	Description
username	Username
primaryEmail	Primary email address
primaryPhone	Primary phone number
name	Display name
avatar	Avatar URL
profile	Standard OIDC profile fields
customData	Additional JSON data for your application

The corresponding patch type is:

type ActionUserPatch = {
  username?: string | null;
  primaryEmail?: string | null;
  primaryPhone?: string | null;
  name?: string | null;
  avatar?: string | null;
  customData?: Record<string, JsonValue>;
  profile?: {
    familyName?: string;
    givenName?: string;
    middleName?: string;
    nickname?: string;
    preferredUsername?: string;
    profile?: string;
    website?: string;
    gender?: string;
    birthdate?: string;
    zoneinfo?: string;
    locale?: string;
    address?: {
      formatted?: string;
      streetAddress?: string;
      locality?: string;
      region?: string;
      postalCode?: string;
      country?: string;
    };
  };
};

Fields such as user ID, suspension state, identities, roles, organizations, MFA configuration, password hashes, and other internal fields are rejected.

For updates, profile and customData are shallow-merged with the existing objects. Returning a nested object with an existing top-level key replaces the value at that key; it is not a deep merge. Identifier updates must also pass Logto's uniqueness checks.

Test context and dry runs

The Test context is sample JSON used only when you click Run test. It is saved with the Action for future tests, but production executions always use the real authentication event.

A dry run:

• Uses the current unsaved script, sample event, and environment variables.
• Executes the script and displays its raw return value.
• Does not save the Action or create or update a Logto user.
• Does not emit a production Action audit event or execution metric.
• Does not apply the production event and result validation for the selected action type.

주의:

A successful dry run proves that the script executed, but not that its result will be accepted during a real authentication flow. Test the complete flow in a non-production tenant before enabling the Action in production.

Successful test results are displayed as returned. Never return passwords, environment variables, API tokens, or other secrets from a script.

Handle errors

The On script error setting applies to execution failures such as a thrown exception, a rejected promise, a failed external request, or a runner failure. It does not make an invalid result valid.

Action type	block (default)	allow
Post first-factor verification	Reject the invalid local credentials.	Not available in Console. Even if set through the API, an execution failure still rejects the credentials.
Post sign-in	Fail the sign-in.	Continue the sign-in without applying the Action update.

For Post sign-in, a malformed or unsupported return value always fails the sign-in, even when allow is selected. Validate every success path in your script and return an explicit supported result or no-op.

Monitor executions

Production executions create independent audit log events:

• Action.PostFirstFactorVerification
• Action.PostSignIn

The audit record includes safe execution metadata such as the Action type, runtime location, duration, decision, and error-policy outcome. Passwords, environment-variable values, script source, and other sensitive values are redacted.

Configure Actions with the Management API

You can also manage Actions through the Logto Management API:

Method	Endpoint	Purpose
GET	/api/configs/actions	List configured Actions
GET	/api/configs/actions/{actionType}	Get one Action
PUT	/api/configs/actions/{actionType}	Create or replace an Action
PATCH	/api/configs/actions/{actionType}	Partially update an Action
DELETE	/api/configs/actions/{actionType}	Delete an Action
POST	/api/configs/actions/test	Dry-run a script with a sample event

The action type values retain their original identifiers for backward compatibility:

• inlineHook.postFirstFactorVerification
• inlineHook.postSignIn

An Action configuration has this shape:

type JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue };

type ActionConfig = {
  script: string;
  environmentVariables?: Record<string, string>;
  contextSample?: JsonValue;
  enabled?: boolean;
  onExecutionError?: 'block' | 'allow';
};

When using the API, set enabled: true explicitly to run the Action. If onExecutionError is omitted, it defaults to block.