[BUG] A potential bypass for token usage within Claude Cloud, utilizing custom agent initialization.

Resolved 💬 1 comment Opened Mar 19, 2026 by notyesbut Closed Mar 22, 2026

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?

This likely involves utilizing session initialization within the Claude code cloud environment, resulting in the initialization of agents that are not subject to the account's usage limits.

What Should Happen?

At the very least, let's verify this. For security reasons, I do not wish to publicly disclose the details regarding the initialization of agents within Claude Cloud. However, I am willing to grant access to a private repository so that you can see for yourself exactly how the agent initialization process is carried out.

Error Messages/Logs

Steps to Reproduce

I believe this may have security implications involving cloud execution, agent/session initialization, and potential unintended billing/isolation behavior.
I have a reproducible setup and supporting evidence, but I do not want to disclose technical details publicly.
Please provide a private security contact or preferred responsible disclosure channel, and I will share the reproduction materials there.

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

claude code windows app

Platform

Other

Operating System

Ubuntu/Debian Linux

Terminal/Shell

Other

Additional Information

_No response_

View original on GitHub ↗

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