Disclosed weakness · no CVE recorded

Claude Code could run an approved command after the user pressed Stop

Anthropic · Claude Code · cloud and self-hosted runner sessions

First documented disclosure · Review checkpoint

Summary

A command could run after Stop when an earlier approval arrived following a restart.

Severity and exploitation evidence

No CVSS score recorded in this review.

Exploitation not established: The reviewed sources describe a correction or controlled reproduction; malicious exploitation against users is not established.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

A delayed approval was applied despite the intervening stop state in sessions without loaded feature flags.

Conditions required

A cloud or self-hosted runner session restarts, feature flags are not loaded, and approval arrives after the user stops the operation.

Potential impact

The previously approved command could execute after the user cancelled it.

Affected versions and configurations

The reviewed upstream report does not establish a complete affected-release range.

Fix and mitigation status

Anthropic lists the correction in 2.1.327.

Practical next steps

Upgrade to 2.1.327 and verify the status of commands stopped during earlier restart or approval races.

Operating lesson

Cancellation must invalidate delayed authority as well as active execution.

Evidence and limits

Anthropic’s release notes are the issue-specific public evidence; the proprietary implementation was not independently inspected. No CVE, score or full affected-version range is provided.

Sources