Remote trigger sessions run indefinitely with no way to stop them

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?

When running a scheduled remote trigger (CCR session), the agent gets stuck in an extended "thinking" loop and never executes any tool calls. The same task completed in under 1 minute
when run locally in Claude Code, but the remote session has been running for 5+ hours with no progress.

Critical issues:

1. No stop button — there is no way to terminate a running remote session from the UI at claude.ai/code/scheduled
2. No hard timeout — the session runs indefinitely, which creates an open-ended billing risk for users
3. Extended thinking loop — the remote agent appears to get stuck thinking without ever calling any tools, even when the prompt explicitly says "do not plan, execute immediately"

What Should Happen?

Expected behaviour: Sessions should have a maximum runtime limit and there should always be a visible stop/cancel button in the UI.

Error Messages/Logs

Steps to Reproduce

• Create a remote trigger with a multi-step prompt (bash + Python + WebSearch)
• Run it manually
• Session enters "thinking" state (shows messages like "germinating", "eludicating", "smooshing") and never progresses

Claude Model

Sonnet (default)

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

 2.1.114 (Claude Code)

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Windows Terminal

Additional Information

_No response_