[BUG] Claude Agent in Xcode 26.4 does not stop to verify user is willing to start using PAYG billing

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

In the Claude Agent provided within Xcode 26.4 and using a paid plan eg. Claude Pro and executing a prompt whilst nearing the 5-hour usage limit (eg. 95-99%), Claude transitions immediately into consuming PAYG extra usage without pausing to verify the user is willing to start using their PAYG credit.

This is in contravention to the online documentation regarding usage limits and gaining extra usage:

"How extra usage works

When you reach your plan’s usage limit with extra usage enabled:
You’ll see a notification that you’ve reached your session usage limit.
If extra usage is enabled and you have funds available, you can choose to continue working.
Your subsequent usage will be billed at standard API pricing rates.
Your session limits reset every five hours as usual."

(https://support.claude.com/en/articles/12429409-manage-extra-usage-for-paid-claude-plans accessed 8th April 2026).

The Claude agent executing within Xcode does not stop mid-processing an instruction and immediately begins consuming PAYG credit without providing any notification the session limit is reached or providing a prompt for the user to continue working.

This has already been reported to Apple's Feedback Assistant and they have completed their investigation saying "Works as currently designed".

What Should Happen?

The Claude Agent should pause with a prompt to the user eg. "Session limit reached. You have enabled extra usage. Continue using PAYG credit?" and expect a Y/N answer.

Error Messages/Logs

Steps to Reproduce

See the description.

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.14 (Claude Code)

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Other

Additional Information

In addition to the reported issue, if extra usage is not enabled, when the 5 hour usage is completely consumed whilst processing a user prompt, the Claude Agent within Xcode 26.4 simply stops. No message is returned to the user indicating what has happened or how long the user should wait. The agent seems completely devoid of many features available from its CLI.

In fact the agent within Xcode is completely unable to tell the user what their current usage whilst within Xcode. Typing eg. /usage at the "Message Claude Agent" in-app prompt produces "Your request couldn't be completed", and attempts at "usage" or "claude usage" cause the agent to try and examine the current Xcode project.

If this is an issue between the Claude agent embedded and distributed within Apple's Xcode app and the user interface to it exposed to the user by Xcode, then Anthropic and Apple need to start having words: the user is not receiving the experience promised by Anthropic.