Cursor Desktop sandbox escape via Claude hook configuration

Summary

Cursor Desktop could execute workspace-defined Claude hook commands from .claude/settings.local.json without dedicated user approval.

Impact

A malicious workspace or agent-created file could configure hooks that run local commands in the user's context when an agent turn ends. This could allow sandbox escape, persistence across turns, local data access, or follow-on compromise.

Remediation

Update Cursor Desktop to a patched version. Workspace-sourced hook commands now require appropriate approval and are subject to the same execution policy controls as other agent shell commands.