sohzm/cheating-daddy
Cheating Daddy
Ce projet est un assistant de réunion IA et un copilote d'entretien qui surveille l'audio système et le contenu de l'écran pour générer des réponses en temps réel lors d'appels vidéo. Il fonctionne comme un outil de transcription audio système et un gestionnaire de prompts sensible au contexte, injectant des documents utilisateur et des profils comportementaux dans les prompts de grands modèles de langage (LLM) pour adapter les sorties de l'IA.
Le système dispose d'une surcouche d'écran furtive, utilisant une fenêtre transparente qui affiche des informations par-dessus d'autres applications tout en restant invisible aux logiciels de partage d'écran et aux outils de surveillance. Il emploie un mécanisme de masquage de processus pour contourner les moniteurs système et rester indétecté.
L'application capture les données d'écran en temps réel et convertit les flux audio système en texte pour le traitement par l'IA. Elle inclut un système de déclenchement par raccourcis clavier pour une opération discrète et une gestion de profil basée sur l'état pour basculer le comportement de l'IA selon le scénario de réunion actif.
Features
- Real-Time Interview Copilots - Analyzes live screen and audio streams to deliver real-time contextual guidance during video meetings.
- Meeting and Productivity Assistants - Provides an AI-powered assistant that monitors audio and screen data to deliver real-time guidance during video calls.
- Copilots - Acts as a real-time copilot that augments user performance during interviews using live audio and visual analysis.
- Audio Transcription - Implements a pipeline that converts live system audio into text for real-time AI processing.
- Real-Time Transcription - Converts live system audio streams into text in real-time to provide a continuous data stream for AI analysis.
- Audio Content Analyzers - Processes real-time system audio to transcribe speech and provide a data stream for AI analysis.
- Personal AI Assistants - Incorporates personal documents and notes to personalize AI context and tailor responses.
- Contextual Prompt Enrichers - Enriches AI prompts by injecting personal documents and behavioral profiles to tailor responses to specific scenarios.
- AI-Powered Answer Generation - Generates automated answers based on captured visual and audio data during live calls.
- Multi-Modal Prompt Injection - Combines real-time audio, visual data, and user documents into multi-modal prompt pipelines.
- Screen Capture Extraction - Captures screen regions in real-time to extract visual data for automated AI analysis.
- Screen Capture Tools - Implements real-time capture of screen content to analyze visual information for AI-generated responses.
- System Audio Capture - Captures internal system audio output to provide a text stream for AI analysis.
- Real-Time Transcription Pipelines - Implements a real-time pipeline that captures raw system audio and converts it to text for AI processing.
- Stealth Overlays - Displays a transparent information window that remains hidden from screen-sharing and proctoring tools.
- Interview Answer Assistants - Analyzes live interview audio and visuals to provide automated answer suggestions during professional screenings.
- Frame Buffer Captures - Extracts raw pixel data directly from the OS frame buffer to analyze visual screen content.
- System Process Hiding - Employs system-level process-hiding techniques to remain invisible to monitoring and proctoring software.
- Prompt Context Aggregators - Aggregates user documents and behavioral profiles to enrich LLM prompts for context-aware responses.
- Behavioral Profile Aggregators - Injects personal documents and behavioral profiles into prompts to tailor AI responses for specific scenarios.
- Screen-Sharing Concealment - Sets OS window attributes to remain invisible to screen-sharing and proctoring tools.
- Transparent Overlays - Renders a transparent, click-through overlay to display AI guidance on top of other applications.
- Global Hotkey Triggers - Provides system-wide keyboard shortcuts to trigger captures and AI responses discreetly.
- Configuration Profile Switching - Toggles between different configuration profiles to instantly change AI context and behavior.
- Interaction Mode Profiles - Manages predefined interaction modes to tailor AI responses to the specific type of meeting.
- OS-Level Input Hooks - Uses low-level OS input hooks to intercept keystrokes regardless of the active window focus.
- Scenario-Based Behavior Profiles - Manages predefined configuration sets to optimize AI behavior for interviews or sales calls.
- Behavioral Mode Switching - Switches between distinct operational modes to adjust AI perception and task prioritization.
- Keyboard Shortcuts - Allows users to trigger AI responses and screen captures using discreet keyboard shortcuts.
