SCOR Hackathon: Automated AI Agent Framework

---

We provide a solution to make better the prompt user requests at SCOR.

• Problem Statement
• Idea / Solution
• Getting Started
• Setup
• Structure
• Authors
• Acknowledgments

🧐 Problem Statement

The use case is the following : at SCOR, employees wish to generate really nice prompts autmated by a framework. So far, prompts were been human written so to extract data from complex and toughtful tracts. Here below our vision to facilite this really time-consuming and repetitive process.

💡 Idea / Solution

()

🏁 Getting Started

To run the project, you need to have the following Python libraries installed:

• PyMySQL - for MySQL database interactions
• Streamlit - for building the user interface

🧭 Setup

1. Clone the repository to your local machine:

   git clone git@github.com:ArthurDelf/Hackathon.git

2. Navigate into the project directory:

   cd Hackathon

3. Install the required Python libraries:

   pip install pymysql streamlit

4. You're ready to go! Simply run the project by using the following command:

   streamlit run main.py

The following AWS services are already set up for you:

• Amazon Lightsail for database storage
• MySQL for database queries
• AWS Lambda for prompt refining
• Amazon Bedrock LLM tuning

All you need to do is run the project, and it will seamlessly interact with these services.

🔨 Structure

├── README.md
├── database
│   ├── __init__.py
│   ├── __pycache__
│   │   └── db_utils.cpython-313.pyc
│   ├── db_utils.py
│   ├── delete_prompt.py
│   ├── export_db.py
│   ├── fetch_prompts.py
│   ├── init_db.py
│   ├── insert_final.py
│   ├── insert_mutant.py
│   └── insert_prompt.py
├── database.json
├── handler
│   ├── Extract_anotations.py
│   ├── OG_handler.py
│   ├── Orignal-prompt-d6cmw.py
│   ├── Orignal-prompt-w6b9q.py
│   ├── action_group_quick_start_jcz36-092x8.py
│   ├── action_group_quick_start_jcz36-3g3rk.py
│   ├── action_group_quick_start_jcz36-56aaj.py
│   ├── action_group_quick_start_jcz36-6tqzz.py
│   ├── action_group_quick_start_jcz36-6z8mb.py
│   ├── action_group_quick_start_jcz36-f0ynb.py
│   ├── action_group_quick_start_jcz36-fh1d7.py
│   ├── action_group_quick_start_pr7p8-6cbwd.py
│   ├── action_group_quick_start_pr7p8-njcxf.py
│   ├── action_group_quick_start_pr7p8-phrgz.py
│   ├── connection_bdd.py
│   ├── main.py
│   └── mutated-prompt-6jiv5.py
└── images
    └── affiche.png

✍️ Authors

• @scartriche - Clement Morice - ECE Paris
• @ArthurDelf - Arthur Delfosse - ECE Paris
• @gfranque42 - Guillaume Franque - 42 Paris
• @replace_stan - Stanislas Barrage - 42 Paris
• @replace_lucas - Lucas Lorang - ECE Paris
• @ivnkla - Kalvin Pichon - Université Grenoble

🎉 Acknowledgments

We are grateful to :

• Antoine & Amar from SCOR
• Leopold from AWS