1Password integrates with Anthropic’s Claude to automate tasks using stored credentials

1Password announced a browser integration for Anthropic’s Claude that enables the AI assistant to use stored credentials—such as usernames, passwords, and API tokens—to automate multi-step tasks like booking travel or managing online accounts on behalf of users.

The feature relies on a new “zero-exposure security framework” developed by 1Password, which injects credentials through a secure channel that the Claude agent cannot view. As a result, the AI can be granted permission to use stored details without exposing passwords, multi-factor authentication codes, or other sensitive information to Anthropic’s models.

Access to credentials is granted on a per-task basis, and users must approve or deny each request with a single biometric prompt. 1Password also scans the page after every autofill to ensure no form submissions retain exposed data before control is returned to Claude.

According to 1Password, the integration locks down access automatically the moment an AI agent takes control of the browser, limiting access to only the credentials explicitly granted for the current task. The company states that nothing else in the 1Password vault is reachable during these sessions.

The feature is available now for 1Password users on Mac, and requires the 1Password desktop app, browser extension, and both the Claude desktop app and browser extensions. 1Password supports storing a variety of security information in its vaults, including passwords, passkeys, 2FA codes, API tokens, and personal details such as addresses and financial information.

1Password did not specify which credentials Claude can access at launch, but indicated that support for payment cards and identity details would be available after launch.