Summary
On July 16, 2026, 1Password launched an integration that lets Claude complete browser logins across the Claude desktop app, Claude in Chrome, Cowork, and Claude Code without exposing usernames or passwords to the model. Credentials are injected into the page at runtime after named-purpose, biometric consent.
What changed
1Password shipped a zero-exposure credential integration for Claude: the user grants named-purpose consent with biometric approval, 1Password injects the login or one-time passcode directly into the webpage (the agent never receives it), and access is verified after fill and is task-scoped. It is Mac-only at launch, covers logins and OTPs only (no cards or identity items), and requires a paid Claude plan.
Why it matters
Credential handling is a core blocker for agentic browsing, since handing an autonomous agent your passwords is unacceptable to security teams. A broker model where the agent acts authenticated but never sees the secret is a plausible template for safe agent authentication, and it puts a major password manager directly in the agent trust path.
Evidence excerpt
The answer isn't handing agents your secrets. It is to let a user give an agent permission to use a credential without letting the agent see it.