Connect 1Password to Claude Code

Inventory the vaults and items a 1Password Connect server can reach, without reading any secret value. Add Toolspoke to Claude Code once and it can use 8 1Password actions. Your 1Password credential stays in Toolspoke, and you choose which actions Claude Code may call.

Set it up in Claude Code

Three steps, the first one once for every toolkit you will ever add.

  1. Add Toolspoke to Claude Code

    Run it in your project, then /mcp to sign in

    claude mcp add --transport http toolspoke https://toolspoke.com/mcp

    The sign-in opens in your browser, and Toolspoke asks you to approve it.

  2. Install 1Password and connect it

    Add 1Password to a project in your Toolspoke workspace.

    Connect server token. You provide it once, when you install the connector. Toolspoke encrypts it at rest and decrypts it only for the length of a single call, and the gateway attaches it to the outbound request itself, so it is never part of the arguments an agent sends.

    Connect server URL
    The root address of your 1Password Connect server, with no trailing path. The API lives under /v1 there, which this connector adds itself. It has to be reachable from this gateway, a Connect server bound to localhost on your laptop is not.
    Connect token
    1Password → Developer → Other → Connect Server → your server → New Token. The vaults you tick while creating the token are the whole of what this connection can ever see, so tick only the ones an agent should be able to inventory, that choice is enforced by 1Password itself and not by anything here. The token is shown once.
  3. Choose what Claude Code may do

    All 8 actions only read, and any of them can be switched off for a project.

What you can ask Claude Code for

A few of the 1Password actions that only read, so they are safe to try first.

  • List the vaults this Connect token can reach, with each one's id, name, description, type and item count
  • Fetch one vault by id, with its name, description, type and item count
  • List the items in one vault, with each one's id, title, category, tags, linked URLs and timestamps
  • Find items in one vault by title or tag, without paging through everything

What Claude Code can do in 1Password

8 actions read.

Reads

  • List vaults

    List the vaults this Connect token can reach, with each one's id, name, description, type and item count. Call it first, every item operation needs a vault id, and the list is also the honest statement of this connection's reach: a vault that was not ticked when the token was issued does not appear here and cannot be reached by any other action. `filter` takes a SCIM-style expression on the name only, e.g. name eq "Staging".

  • Get vault

    Fetch one vault by id, with its name, description, type and item count. Use it to confirm which vault an id refers to before listing what is in it. Vault types are PERSONAL, USER_CREATED, EVERYONE and TRANSFER.

  • List items

    List the items in one vault, with each one's id, title, category, tags, linked URLs and timestamps. This endpoint returns item overviews and never carries field values, so it is the safe way to see what a vault holds: which credentials exist, what they are called, what they are for, and when they were last changed. Categories include LOGIN, PASSWORD, API_CREDENTIAL, SERVER, DATABASE, SSH_KEY, SECURE_NOTE and DOCUMENT. Use search_items to narrow by title or tag rather than paging through a large vault.

  • Search items

    Find items in one vault by title or tag, without paging through everything. `filter` is a SCIM-style expression Connect accepts on those two properties only: title eq "Production database", or tag eq "billing". Like list_items it returns overviews and no field values. This is the operation to reach for when the question is "where is the credential for X", it answers where, and nothing here answers what.

  • Get item metadata

    Read one item's metadata, its title, category, tags, linked URLs, version, and who last edited it and when. It is named metadata rather than get_item because that is exactly what it returns: 1Password's own endpoint answers with the whole item, and this operation strips `fields`, `sections` and `files` out of the response inside the gateway, before an agent sees it and before anything is written to the audit log. There is therefore no password, no token, no private key and no note body in the result, and no argument to this operation can put one there. If a value genuinely has to be used, a person retrieves it from 1Password; this connector's job is to say where it lives, not to hand it over. The response is withheld from the audit log as well, so not even the metadata accumulates a second copy.

  • List item files

    List the files attached to one item, with each file's id, name and size in bytes. Names and sizes only: the `content` key is stripped from the response and this connector does not expose the `inline_content` parameter that would make Connect fill it in, so an attached certificate, key file or recovery-code document is visible as an entry and never as bytes. Useful for answering "is the private key stored alongside this server record" without retrieving it.

  • List activity

    Read the Connect server's own request log, which token made which request, against which vault or item, when, from which address, and whether it was allowed. This is how to answer "what has actually been read through this Connect server", including by this connection itself. Page with `limit` and `offset`. Withheld from the audit log because each entry names an actor and the address they called from.

  • Get health

    Report the Connect server's own state, its name, version, and the status of each service it depends on, such as the sync process and its local database. Use it when a call fails to tell "the Connect server is unwell" apart from "the token cannot reach that vault". It is served at the server root rather than under /v1, and it says nothing about any vault or item.

1Password in Claude Code: questions

Does Claude Code see my 1Password credential?

No. The credential is stored encrypted in Toolspoke, and the gateway adds it to each call. Claude Code only ever talks to Toolspoke, and the dashboard shows the last four characters of the credential and nothing more.

Will 8 actions crowd Claude Code's context?

No. Toolspoke shows Claude Code five bridge tools, plus a guide where a toolkit has one, and Claude Code searches for the 1Password action it needs when it needs it. Adding more toolkits does not add more tools to its list.

Can I add it once for every project?

Yes. Add --scope user to the command and Claude Code keeps Toolspoke in every project on this machine. Without it, the server is added to the project you ran the command in.

What does it cost?

Tool calls are free on every plan, and a 1Password call does not run in a sandbox, so it bills nothing.

Try 1Password in Claude Code

Free to start, and tool calls cost nothing on any plan.