CodexDataLab is a terminal-first, human-in-the-loop “chat with your data” workspace for small-to-medium tabular datasets. It uses a Textual TUI for fast iteration and integrates Codex to help users ingest data, clean it, explore it, visualize it, and produce reproducible outputs.

See VISION.md for the project vision, goals, and roadmap.

Install the latest release with a single command:

curl -fsSL https://raw.githubusercontent.com/SzymonPrajs/CodexDataLab/main/scripts/install.sh | bash

Once installed, launch the application:

codexdatalab
# Or use the short alias:
codala

Prerequisites: Python 3.10 or higher.

The script installs the application to ~/.local/share/codexdatalab and links binaries in ~/.local/bin. If the command is not found after installation, ensure your local bin is in your path:

export PATH="$HOME/.local/bin:$PATH"

Nightly Builds To install the latest changes from the main branch, use the nightly flag:

curl -fsSL https://raw.githubusercontent.com/SzymonPrajs/CodexDataLab/main/scripts/install.sh | bash -s -- --nightly

To set up the project locally for contribution or modification:

# Create and activate a virtual environment
python -m venv .venv
source .venv/bin/activate

# Install dependencies in editable mode
pip install -U pip
pip install -e .

# Run locally
codexdatalab