CSV AI, Streamlit app

Requirements

This is a legacy local experiment pinned to 2023-era packages. Use Python 3.11 for the existing pandas pin; Python 3.12+ is not validated. A current provider model and a dependency migration are needed before considering hosted use.

Setup

python3.11 -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements.txt
export OPENAI_API_KEY="your-key-here"
streamlit run app_csv.py --server.address 127.0.0.1

Set your key in the shell environment. Never paste it into source code or commit it. .env.example documents the variable; .env files are not loaded automatically.

Upload a UTF-8 CSV, enter a question, and click Ask. Creating the model and agent happens only after a question and key are available.

Scope and limitations

Use trusted, non-sensitive CSVs in an isolated local environment. Questions and CSV context can be sent to OpenAI. The legacy LangChain dataframe agent can execute generated Python with the app process's permissions; it is not a sandbox and this app is not suitable for public uploads. See LangChain's discussion of agent code execution.

The audit checks the UI control flow offline; it does not establish live OpenAI model availability or make this old dependency stack production-ready.

python -m unittest discover -s tests -v

The author

Hi 👋 I'm Pierre-Henry Soria. A super passionate and mission-driven software engineer! 🤠 I'm also a true cheese 🧀, dark chocolate, and espresso lover! 💫