In recent years, the proliferation of HID attacks has raised significant security concerns. Previous research has primarily focused on detecting these attacks by analyzing behavior features extracted from keystroke timestamps. However, the sophisticated HID attack could highly imitate user behavior features with the rising trend of countermeasures to overcome the known flaws of existing studies. In this paper, we propose a novel scoring algorithm that relies exclusively on keystroke text analysis. It depends on a HID keystroke text feature space, within which each feature matching result of a text segment is scored and their sum is calculated. Once the total score surpasses a predefined threshold, the text segment will be flagged as indicative of an unsafe HID interaction. In confrontational testing, our algorithm demonstrates a precision rate exceeding 95% in detecting HID attacks.