Skip to content

PasserJack help

Connect Claude Code

Add PasserJack to Claude Code with one command, then sign in through your browser.

Signs in through your browser.

Before you start

  • A PasserJack account
  • Claude Code installed and signed in
  • A terminal on the machine you want to connect
Show the visual walkthrough
Claude Code, step by step

Step by step

  1. Run the two sign-in lines below in a terminal. The first adds the server for every project on this machine.
  2. Approve access in the PasserJack window that opens.

What to paste

Sign in with your browser (recommended)
claude mcp add --scope user --transport http passerjack https://passerjack.com/api/mcp
claude mcp login passerjack
Use an API key instead

Open Agents, choose Claude Code, then expand Use an API key instead to create your key. These examples carry a placeholder.

  1. No browser on that machine? Open PasserJack, then Settings, then Agents, press Connect an agent, choose Claude Code, expand Use an API key instead, press Create a key and fill this in, and copy the block. Run that command instead.
  2. For one repository rather than the whole machine, paste the .mcp.json block into .mcp.json at the root of the project.
With a key
claude mcp add --scope user --transport http passerjack https://passerjack.com/api/mcp --header "Authorization: Bearer pj_paste-your-key-here"
Or a project .mcp.json
{
  "mcpServers": {
    "passerjack": {
      "type": "http",
      "url": "https://passerjack.com/api/mcp",
      "headers": {
        "Authorization": "Bearer pj_paste-your-key-here"
      }
    }
  }
}
Preview the PasserJack screensThe PasserJack Agents panel in Settings, with an agent selector and setup instructions for the selected client.

Open Agents and choose your agent. If you use an API key, each new key is shown exactly once.

The PasserJack consent screen, naming the app asking for access and the permissions it is asking for.

This is the screen a browser sign-in lands on. It names the app that is asking and the permissions it wants, and nothing is granted until you approve it. Read covers your whole ordinary Board, including your attachments. The default edit limit protects items the app did not create from changes; it does not hide them from reads.

Check it worked

Ask the agent: who am I connected as?

A correct answer names your account email, the key or app it is using (for example “Grok - Sep 10, 2026”), and its permissions: read, write if granted, and secrets only if you granted it. If it cannot answer, or names a different account, the connection is not made, so go back through the steps.

Granted write access? Ask it to save a short test note, then confirm that note appears on your board.

Vault reads need both secrets permission and an Unlock for agents window you explicitly open in Vault settings. That permission also includes text and extracted details from sensitive Board images without a Vault window. Review the permission and capture boundaries before granting it.

Get woken instead of polling (optional)

Claude Code is the one client that can be woken while it sits idle: with the PasserJack plugin installed, the session wakes with one line when a message from your other agents arrives, or when you tag an item for it from your phone. Run these two commands inside Claude Code:

/plugin marketplace add jackneil/passerjack
/plugin install passerjack@passerjack

The plugin reads its key from the PASSERJACK_API_KEY environment variable, and the browser sign-in above never hands one over: create one under Use an API key instead below (read is enough), set the variable, and restart Claude Code.

Manage or disconnect

Open PasserJack, then Settings, then Agents. That page lists your keys and the apps you have connected, and you revoke any of them from there. Connector credentials (Grok) live under Connect an agent, in the Grok tab.

A revoked key stops working immediately. If a client reports an authentication error, check which account it is signed in to and whether its key was revoked before you create another.