How github-harsha63/genai-ethical-evaluation-game works
The product is an interactive ethical-decision simulation: users answer randomly selected dilemmas, submit their choices, and receive generated feedback, with options to retake or start a new test. The Streamlit app calls a Flask API; the sampled backend uses Gemini, despite the README’s Mistral description. The separate backend/mistral.py also calls Gemini but is not shown as wired into the Flask app. The graph reflects only evidenced calls and documented behavior; question selection and feedback generation are represented as stages within the backend file that implements them.
Main entry points: app.py Simulation app · main.py Flask API · mistral.py Separate AI helper · images Scenario images
Components 14
User experience
- Simulation app
frontend/app.py - Scenario display
frontend/app.py - Choice tracking
frontend/app.py - Feedback and retake
frontend/app.py - Scenario images
frontend/static/images
Simulation API
- Flask API
backend/main.py - Question pool
backend/questions.json - Random selection
backend/main.py - Response submission
backend/main.py - Feedback prompt
backend/main.py
AI feedback
- Gemini request
backend/main.py - Separate AI helper
backend/mistral.py
Other components
- Participant
- Gemini API
Connections 18
- Participant → Simulation app: uses
README.md - Simulation app → Flask API: requests questions
frontend/app.py - Flask API → Random selection: routes to
backend/main.py - Random selection → Question pool: samples from
backend/main.py - Simulation app → Scenario display: displays
frontend/app.py - Scenario display → Scenario images: loads
frontend/app.py - Scenario display → Choice tracking: records choice
frontend/app.py - Participant → Choice tracking: chooses
README.md - Choice tracking → Simulation app: submits responses
frontend/app.py - Simulation app → Flask API: posts responses
frontend/app.py - Flask API → Response submission: routes to
backend/main.py - Response submission → Feedback prompt: builds prompt
backend/main.py - Response submission → Gemini request: requests feedback
backend/main.py - Gemini request → Gemini API: calls
backend/main.py - Response submission → Flask API: returns feedback
backend/main.py - Simulation app → Feedback and retake: shows feedback
frontend/app.py - Feedback and retake → Participant: presents result
README.md - Separate AI helper → Gemini API: calls
backend/mistral.py