claude-code-pro

This is a task-launching automation script with no clear malicious payload. Its primary security concerns are intentional execution of Claude Code with permissions skipped in auto mode and shell command injection through the unvalidated --model value because CLAUDE_CMD is sent to an interactive shell as text. Use plan mode by default, validate or safely quote model input, and avoid running it in sensitive workspaces.

The provided fragment is a README for an automation workflow and contains no executable implementation or direct evidence of malware. The primary identified risk is the documented auto mode, which runs Claude Code with --dangerously-skip-permissions and can permit unrestricted changes or commands in the target directory. The referenced scripts and package contents must be reviewed before installation or use, particularly for shell injection, unsafe workdir handling, and unexpected network or system actions.