Hermes Agent 1Password Integration: How It Works (2026)

Wondering how the new Hermes agent 1Password integration actually works? The quick version is just below. The slightly longer version: with v0.21.2, released on 11 September 2026 per the official GitHub release notes, Hermes Agent gained a password-blind credential vault that can sign into sites, pay and fill addresses from 1Password, Bitwarden or the local Hermes vault, without the agent ever seeing a secret. I’ve wanted exactly this since the first time I watched an agent stall at a login screen.

How the Hermes Agent 1Password integration works

The design principle is the interesting part. Rather than handing your agent a plaintext password, the vault is password-blind: credentials stay inside 1Password, Bitwarden or the local Hermes vault, and the fill happens without the secret ever entering the model’s context. The v0.21.2 release notes (tracked as PRs #106480 and #107585) put it plainly: the agent can “sign in, pay, and fill addresses from 1Password, Bitwarden, or the local Hermes vault without ever seeing a secret”.

Why that architecture matters for anyone running agents seriously:

• No secrets in transcripts. If the model never sees the password, it can’t leak into logs, session exports or model context.
• Real workflows unlock. Sign-ins, checkouts and address forms are exactly where browser-driving agents used to stop and wait for a human.
• 2FA is handled sanely. Per the release notes, two-factor codes come from saved authenticator keys, or the agent prompts you for entry, so you keep 2FA on everything.
• It extends to plugins. Private git plugins can install using stored credentials, which quietly fixes a real team-workflow annoyance.

Hermes Agent 1Password integration setup: step by step

Here’s the honest setup path, based on the official release notes:

1. Update Hermes. The vault landed in v0.21.2 (11 September 2026); the current tagged release is v0.21.3 (14 September 2026). Existing installs: run hermes update in the terminal. Docker users: images are built from the release tag.
2. Pick your credential source. 1Password, Bitwarden or the local Hermes vault are the supported options named in the release notes.
3. Connect it in Hermes’ settings and grant access to the vault entries you actually want the agent to use — least privilege, always.
4. Set up 2FA handling. Save authenticator keys where you want hands-off runs; leave prompted entry for anything sensitive.
5. Test on a throwaway login first. Watch one full sign-in end to end before you point it at anything that can spend money.

The release notes don’t document every settings screen, and beta-fresh features move fast, so treat the in-app flow as the source of truth for clicks and toggles.

What else shipped around the vault

Context helps here, because v0.21.2 wasn’t a one-feature release. Per the GitHub release notes it bundled roughly 312 merged PRs, headlined by a state.db reliability campaign (six root causes of database fragility addressed, 44 issues closed) and a curated, SHA-pinned plugin catalogue. Three days later, v0.21.3 (14 September 2026) rolled up around 338 more PRs, fixing remote dashboard sessions expiring during refresh bursts and duplicate state.db writer handles in long-lived processes.

Release	Date	What it means for the vault
v0.21.2	11 September 2026	Credential vault ships: 1Password, Bitwarden, local vault; password-blind fills; 2FA via saved authenticator keys or prompt
v0.21.3	14 September 2026	Stability rollup on top — the version you should actually install

My take after a week of watching this release line: the vault is the piece that turns Hermes from a clever autopilot into something you can trust with real accounts, and it’s the feature I’d update for first. It’s also exactly the kind of capability we fold into client-facing automation carefully — with scoped vault access and a human review step — which is how we approach it in the Boardroom builds too.

The bottom line on the Hermes Agent 1Password integration

If you run Hermes Agent and you use 1Password or Bitwarden, update to v0.21.3 and switch the vault on: you get sign-ins, payments and form fills inside agent runs without secrets touching the model, and 2FA stays intact. Start with scoped access and a test login, then expand. It’s the most practically useful thing in the v0.21.x line so far. And if you want help deciding what your agents should and shouldn’t be allowed to touch, book a free SEO strategy session and we’ll design the workflow together.

What’s new in Hermes v0.22

Hermes Agent v0.22.0, released on 14 October 2026, made password-manager sign-in a headline feature. The release notes put it simply: the browser agent can log into sites with logins from 1Password, Bitwarden or Hermes’s own encrypted vault, and the model never sees the password.

• A masked prompt when nothing is saved. If there’s no saved login for a site, you get a masked prompt in the CLI, TUI or Desktop to save it, so you type the password into Hermes, not into the chat.
• One-time codes stay out of the chat. Per the notes, one-time codes are filled without entering the conversation.
• Cards and addresses need your OK. Password-manager sign-in can also fill cards and addresses, but card and address fills at checkout need your confirmation first.
• Vault tools load on demand. The vault tools now load only when they’re needed, part of a change that trims about 2.1k prompt tokens from each request.
• Owner-only files at rest. Credentials, state.db, snapshots, browser profiles and media caches are now stored as owner-only files whatever your umask.
• The Settings toggle works again. Detected password managers can be switched on from Settings again; the notes say the toggle had been a no-op.

If you set up the Hermes Agent 1Password integration on v0.21, update to v0.22 to get all of this. Your existing vault setup carries over.

Want another angle? Read the AI Profit Boardroom guide to Hermes 1Password.

FAQ: Hermes Agent 1Password integration

When was the Hermes Agent 1Password integration released?

It shipped in Hermes Agent v0.21.2 on 11 September 2026, per the official GitHub release notes, as part of a new password-blind credential vault, and rolled forward into v0.21.3 on 14 September 2026.

Does the Hermes agent actually see my 1Password passwords?

No. The release notes describe the vault as password-blind: the agent can sign in, pay and fill addresses from 1Password, Bitwarden or the local Hermes vault “without ever seeing a secret”.

Does it support Bitwarden too?

Yes. The v0.21.2 release notes name 1Password, Bitwarden and the local Hermes vault as supported credential sources.

How does two-factor authentication work with it?

Per the release notes, two-factor codes come from saved authenticator keys, or the agent prompts you to enter the code yourself when one is needed.

How do I get the credential vault on my install?

Update to v0.21.2 or later. Existing installs can run the hermes update command in the terminal; v0.21.3 (14 September 2026) is the current tagged release and also ships as Docker images built from the tag.

What changed for the 1Password integration in Hermes v0.22?

Hermes v0.22.0 (14 October 2026) lets the browser agent log into sites with 1Password, Bitwarden or the Hermes vault without the model seeing the password, adds a masked save prompt when nothing is saved, fills one-time codes without them entering the conversation, and asks you to confirm card and address fills at checkout.

Related reading

• How to update Hermes Agent
• Hermes Agent state.db corruption fix
• Hermes Agent review

Last updated September 2026. This is the living guide to hermes agent 1password integration — it gets updated as the tools change.