This repository is dedicated to the SNITCH capstone project for the Spring 2024 graduating class at Southwestern University.
This tool is meant to be useful in detecting the use of AI on homework/project submissions. If you are a student, you will be able to view how much your work matches the output of different AI models. To avoid the temptation of using any AI generated work, the student view is limited to only being able to see a match percentage. For admins (most likely teachers or professors), a different view will be available, where AI generated output will be avaialable to see and compare to the submission (for different AI models).
The anticipated use of this tool is to utilize various APIs, some of which require a secure API Key to be generated. To do this, look at the file called .gitignore. This file specifies which files should be ignored when it comes to pushing those files to this public repository. If you use an API key in the actual files in this repository, then push those files, the API key may become invalid (and will no longer work). To avoid this from happening (and to practice good security practices), simply create a file in the main project directory named config.py. In that file, you can assign a variable the value of your API key. The main project file, API.py, imports config.py, and can access the local keys. If you need further assistance doing this, you can follow this quick reference video.
Example of how an API key can be stored in a variable in the config.py file: api_key = "mykey123"
*Note: This is a fake API key
For this project, multiple python packages will need to be installed. To install them cd snitch-ai and run this command pip install -r requirements.txt
You will also need to have npm installed to run the react app. After that cd snitch-ai and run this command npm install to get all of the node modules needed to run the project.
If there are more packages that need to be installed, you will likely get an error saying that x package could not be found. This means you just need to pip install that package. In the future, we will incorporate a requirements file that can be run instead in order to install all necessary packages.
Finally there are two config files that need to be generated with relevant API keys. One needs to be created in the snitch-ai directory and it needs to be titled config.py. Keys for this file can be found in config google doc that in the shared google drive. The second needs to be created in the python-code directory and also needs to be titled config.py. For simplicity, keys for both of these files are the same and can be found in the config google doc in the shared drive.
This project currently has two components, the React App that serves as what the user will actually see/use, and the FastAPI, which acts as the connection point between several of our APIs and React app.
cd snitch-ai followed by npm start. The first command changes the directory to the react app directory, and the second command is used to run the application. Once this runs, a window should pop up, allowing you to see the main page of the react app.cd python-code . Once you are in this directory, you can run python3 -m uvicorn API:app --reload. This command is what is used to start the API. If you are successful in running this command, then you should see a link to where the API can be found in the terminal. You can copy and paste this link into a browser in order to mess around/test API functions that are found in API.py. This is not necessary to open if you are focusing on the React app. You simply need to have this running in order for the React app to send data to the various APIs we are using (i.e. ChatGPT and Gemini).
Docker stuff:
docker build -t snitch-fastapi-image .docker run -p 8000:8000 snitch-fastapi-image