Skip to main content
An expired OAuth access token and an expired OnlyFans session need different fixes. Refresh the access token through Tokens and refresh. When an OnlyFans account needs a reconnect, create a targeted journey for its owner. Pause actions after the OnlyFans session expires. Ask the owner to reconnect, then read the current account record. Resume with the same account ID only when status is connected; otherwise follow current recovery. Pause only the affected account. A successful token refresh does not change the OnlyFans account’s reconnect status. Confirm current account status before resuming actions.

Know who handles each step

Detect an expired account

Pause actions for the affected linked account when a request returns linked_account_reconnect_required, its status is needs_reconnect, or you receive linked_account.expired. Keep the account ID and authorization. Other linked accounts may still be usable.

Ask the owner to reconnect

For an Express implementation, see the customer integration source example. It includes app sessions, account selection, reconnect routes, and PostgreSQL storage. Create a short-lived journey with the authorization’s current access token:
Generate a random, one-use state and bind it to the initiating app session. redirect_uri must exactly match a callback registered for the OAuth client. Use the returned url rather than constructing a Link URL. The journey expires at expires_at and targets only linked_account_id under authorization_id. If your auth library owns the OAuth callback, it will reject the reconnect return as an unknown OAuth state. Use a separate return URL instead:
  • Developer integration: add the return URL to the integration’s redirect URIs in the dashboard.
  • Operator private app: register it as ofauth_reconnect_return_uri (see Register a reconnect return URL). It is accepted only by reconnect, never as an OAuth redirect URI.
The owner authenticates to OFAuth and reconnects the exact account. The target and return URL cannot be changed during the journey. Reconnect retains the authorization, linked account ID, and scopes. Sandbox sample accounts use the dashboard’s Reconnect action instead of provider sign-in.

Handle the return

Keep reconnect state on your server. Bind each value to the app session, the token-storage key, and the linked account being restored. Create this table once:
Before opening the journey, insert a random state with its bindings. Keep the state for 30 minutes so an expired journey can still return a clear result. The journey itself expires at expires_at in the API response. The return handler below uses the pool and ofauth from Store customer tokens. Pass the current trusted app session from your server. Read the target account from the saved state, never from callback parameters.
currentAccessToken is your auth framework or application’s token getter. Mount this handler at your registered return URL, outside the auth library’s OAuth callback route. If the app session has expired, ask the user to sign in and start a new journey. A callback from another session or a repeated callback cannot consume a valid pending return. If the user selects another account during reconnect, this handler still checks the original target. The handler checks the current account for every valid outcome. Resume actions only when canResume is true. If the lookup fails, keep the account paused and follow the error’s recovery action. An account may still need attention even when the return says completed. The source example includes the Express route, the state insert, and app-session storage. For access-token recovery, see Handle failures.

Resume actions

After you receive linked_account.reconnected, call GET /v3/linked-accounts/{id} with the authorization’s access token. Resume when its status is connected. The linked account ID stays the same. If the account was removed or the authorization revoked, follow the error’s recovery action to request access again. See Manage linked accounts. Before resending an action interrupted by a network failure, check whether it completed. See Handle failed requests.