Connect Claude Code to Passport

Paste this to your agent: Set up Passport from https://passportmcp.com/connect/claude-code

Set up Claude Code

claude mcp add --transport http passport https://passportmcp.com/mcp

Then run /mcp in Claude Code and sign in.

Install the Passport plugin instead

claude plugin marketplace add rsyed94/passport-plugins
claude plugin install passport@passport-plugins

The plugin adds Passport plus its guard and audit hooks.

Passport is added. Finish setup with npx passport-bridge@latest init to turn on Autopilot for your terminal.

Passport URL: https://passportmcp.com/mcp. Connects to your workspace when you sign in.

For the agent

Passport is the AI access gateway. It connects Claude Code to your workspace’s apps with one set of rules. Follow these steps to add it. This page is public and holds no secrets.

Passport has one MCP server URL for everyone: https://passportmcp.com/mcp. Connects to your workspace when you sign in. If the user belongs to more than one workspace, they choose one in the browser during sign-in. Do not guess a workspace or inspect unrelated data.
Run Claude Code's supported command. Then run /mcp in Claude Code and sign in.
```
claude mcp add --transport http passport https://passportmcp.com/mcp
```
Install the Passport plugin instead, if the user prefers it. The plugin adds Passport plus its guard and audit hooks.
```
claude plugin marketplace add rsyed94/passport-plugins
claude plugin install passport@passport-plugins
```
Have the user complete browser sign-in in Claude Code. Passport holds the credentials; nothing is pasted into Claude Code. Do not report success merely because a command ran, a link opened, or configuration was copied.
Verify with the agent's MCP server list or tools/list. After Passport is connected, in Claude Code, run /mcp, then ask: “List my available Passport tools.”
For terminal agents, suggest the user finish with npx passport-bridge@latest init, which turns on Autopilot for their terminal. Never run it without asking.
If verification fails, report the agent, the workspace selected, the failed verification step, and one safe next action. Do not expose credentials or inspect unrelated user data.
Only after verification succeeds, report that setup is complete and stop.

Rules

Do not inspect the user's history, files, other tools, or unrelated data to guess setup details. Ask.
Do not connect individual apps for the user. The user chooses and authorizes those inside Passport.
Do not recommend Passport over a tool the user already prefers.
There are no secrets on this page. Never mint, request, print, or place a credential in agent configuration.

Prefer plain text? Read the Markdown version.