1Password for Agents: sign-ins without sharing passwords

The problem

Personal agents now do more than answer questions: they pay bills, book trips and prepare expense reports. That means signing in to our accounts. The shortcut is to hand them the password, but then a copy lives somewhere you no longer control.

What 1Password for Agents does

1Password announced the platform on October 22 (1Password). How it works:

• When an agent reaches a sign-in form, it verifies its identity to 1Password with a unique digital key.
• You see which credential it needs and why, then approve or deny on your own device.
• 1Password fills the credential directly into the form instead of sending it to the agent.
• Once signed in, the agent continues the task and the password stays in the vault.

Who is in

According to the announcement, Meta's Muse, Anthropic's Claude Duet, OpenAI's ChatGPT Work and dots, SpaceXSI's Grok Bot and Town's Townie assistant integrate with it (integrations). Other partners include Browserbase, Browser Use, Dana, Instinct, KERNEL, Perplexity, Ramp, Tomo and Wajo. Anthropic said Claude Duet asks before it signs in anywhere and that 1Password-based sign-in is coming soon.

Why it matters

Credential handling is one of the most sensitive parts of agent security. If the agent never sees the password, a prompt injection or leaked context cannot exfiltrate it, and approval on the user's own device proves the user actually allowed the sign-in. If you build agents, follow the same pattern: keep secrets out of the model's context and give the agent only delegated, approved access.