1Password for Claude is available now. Claude can finish browser tasks that need logins and one-time codes. 1Password says those secrets never enter the model, its memory, or Anthropic’s systems.
As a result, agent users no longer face a blunt tradeoff. Hand Claude the password, or stop and finish the job yourself. Instead, 1Password stays the source of truth. You grant access only at runtime, with your approval.
How the zero-exposure flow works
When Claude hits a login wall, 1Password shows which credential Claude needs and why. After you unlock with biometrics, 1Password injects the secret into the page. Claude never sees the vault item, password, or one-time code.
Meanwhile, 1Password limits that access to the current task and cuts it off once the task ends. After autofill, 1Password also checks that secrets did not remain on the page. If submission fails, it clears the filled values before returning control.
“We need a new security model that is purpose-built for agents, not just humans,” said Nancy Wang, CTO of 1Password. “The answer isn’t handing agents your secrets. It is permission to use a credential without letting the agent see it.”
Agentic Mode locks the vault under AI control
Alongside the Claude hookup, 1Password is shipping Agentic Mode in the browser extension for all users. When a compatible agent takes over the browser, the extension locks down. Consequently, the agent can only use logins and codes you explicitly approve for the current task.
Importantly, Agentic Mode works even if you never enable the full Claude integration. 1Password says it will expand beyond Claude as more agents arrive. Qualifying enterprise accounts also get the same explicit approval model for work credentials.
Availability and setup requirements
1Password for Claude is live on Mac for business, family, and individual plans. Getting started requires the 1Password desktop app and browser extension, plus the Claude desktop app and Claude in Chrome. In addition, 1Password plans later support for payment cards and identity details after launch.
The launch sits next to a wider agent push. Anthropic is also expanding how people direct agents from phones in Claude Cowork Mobile. OpenAI is racing the same surface with ChatGPT Work. Because of that race, credential control is becoming the real bottleneck, not just model skill.
Still, claims like “never reaches Anthropic” depend on trust in the design. If you already live in 1Password on Mac, this is the cleanest agent-login path on offer today. However, Windows and deeper payment support are the next tests.
