Feature Request: reduce browser-dependent auth friction for Sign in with Apple (Hide My Email) users

Resolved 💬 1 comment Opened Jun 6, 2026 by xwrvnggp9n-star Closed Jul 13, 2026

Preflight Checklist

  • [x] I have searched existing requests and this feature hasn't been requested yet
  • [x] This is a single feature request (not multiple features)

Problem Statement

I created my Claude account with Sign in with Apple using Hide My Email, so the account is keyed to a @privaterelay.appleid.com address.

There is no Sign in with Apple button on the web, and no password option, so my only browser login path is the email magic link. Authenticating Claude Code relies on a browser OAuth flow. When I am already logged into claude.ai in my default browser, the authorize step is one click. But in a fresh or remote environment (new container, clean profile, different machine), I have to run the full magic-link round trip: look up the relay address, request the link, open mail, then click the link or copy a verification code. That is the slow path, and it recurs.

This affects anyone who signed up with Sign in with Apple plus Hide My Email and then uses Claude Code. The relay address, no password, and no web Apple button together mean there is no quick browser login for this group.

Proposed Solution

Any one of these would remove the friction:

  1. A long-lived, subscription-tied auth credential for Claude Code that does not depend on an active browser session. (An API key works today but bills against API credits, which is a different cost model from a Max or Pro subscription.)
  1. Let an account link a second sign-in method (for example Google) so there is a faster browser login than the relay magic link.
  1. Allow an optional account password as a fallback for web and desktop sign-in.

The core need: one quick, repeatable way to authenticate in a browser or a fresh CLI environment that does not require retrieving a privaterelay alias and round-tripping through email every time.

Alternative Solutions

Workarounds I use today:

  • Keep one browser profile permanently logged into claude.ai and never clear its cookies, so the Claude Code authorize step is a single click there.
  • For fresh or remote environments, authenticate Claude Code with a Console API key, which skips the browser but bills against API credits instead of my subscription.
  • Request the magic link and open it on the same device to avoid the cross-device verification-code step.

None of these is a real fix. They reduce how often the magic-link round trip happens, but the account still has no fast, repeatable login for relay-email users.

Priority

Medium - Would be very helpful

Feature Category

Configuration and settings

Use Case Example

Example scenario:

  1. I use Claude as my primary tool every day across a Mac and an iPhone.
  2. My account was created with Sign in with Apple and Hide My Email, so its email is a privaterelay.appleid.com alias.
  3. I spin up a fresh session, or open Claude on a machine that is not my main one.
  4. There is no Apple button on the web, no password, and the relay alias is not something I have memorized, so I look it up, paste it, request a magic link, switch to mail, wait, then click the link or copy a code.
  5. With any of the proposed options, that step becomes one click or one stored credential, and I get back to work.

Additional Context

Related existing issues:

  • #58329 (open): Apple Private Relay addresses treated as a shared organizational domain, causing account lockout and exposing other users' organizations. A more severe symptom of the same relay-email handling.
  • #56951 (closed): Hide My Email magic link hijacked by a Chrome Google session, making the Chrome extension unusable for these users.

This request is the milder, broader version. Rather than a specific lockout, the day-to-day issue is that Sign in with Apple plus Hide My Email leaves no fast browser or CLI login path. A large share of Apple users sign up this way, so this likely affects many people beyond me.

View original on GitHub ↗

This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗