auto-researching

The skill’s capabilities largely match its stated purpose, but it is high-risk because it grants an AI agent indefinite autonomous execution, code modification, git history changes, and command running after minimal user interaction. This looks more like a powerful but risky automation skill than confirmed malware.