Poteto Mode

SUSPICIOUS. The skill is mostly coherent as an agent operating-style skill and the referenced dependencies appear first-party, so there is low supply-chain concern. The main risk is scope: it grants broad autonomous tool use and external actions by default, which is powerful and can have real-world effects even without explicit concealment or credential theft.

No malware or covert data-theft behavior is present because the fragment is an explicit cleanup playbook rather than executable package code. It presents a significant operational data-loss risk through forceful worktree, simulator, runtime, application-state, and cache deletion. Execution should require independent path validation, explicit confirmation for every non-clean or uncertain item, and avoidance of `rm -rf` unless the exact surviving path is verified.