1Password for Agents: Let AI agents sign in for you without seeing your passwords

A supported agent that reaches a sign-in page presents its own digital key to 1Password. The user then sees which login the agent wants and why, and approves or rejects the request on a personal device. After approval, 1Password enters the password or one-time code into the site's form, so the agent completes the login without handling the credential. Connecting an agent does not open up the entire vault. Users choose which login it can use and can disconnect the integration later. Integrations with Meta Muse and Grok Bot are rolling out over the coming days. Claude Duet, ChatGPT Work, dots, and Town's Townie are coming soon. The feature is included at no extra cost with Individual and Family accounts hosted on 1Password.com. Sign-ups are handled through a waitlist. The launch extends a earlier step: the Claude integration introduced in July, which enabled approved browser sign-ins on Mac. Password managers built for browsers were not designed for this use. People who use browser-saved passwords or passwords written in notes have discovered that their agents cannot use those passwords. The main limit is what happens after sign-in. Once signed in, the agent may be able to read information or take actions within that account. Those permissions depend on the agent and the website, so protecting the password is not the same as limiting everything an assistant can do afterward. For builders, permission to log in and permission to act are separate decisions, so connected products will need their own scoping, audit trails and revocation paths.