Python 3.11+
The setup script creates an isolated project environment.
Official download ↗This local edition is not a simulated demo. It starts the genuine Chartsmith workflow, streams each stage into the studio, and keeps its run artifacts on your computer.
The quick path is verified on Windows 10 and 11. A normal ChatGPT sign-in is enough; you do not need to create or paste an API key.
The setup script creates an isolated project environment.
Official download ↗Node.js must include npm. Exact web packages come from the lockfile.
Official download ↗First setup downloads requirements; redesigns call Codex models.
Use a ChatGPT account with Codex access in the browser login flow.
OpenAI auth guide ↗Put it in a normal writable folder. Do not run it from inside the ZIP preview.
In Windows Explorer, right-click the folder and choose “Open in Terminal.”
The first launch can take a few minutes. Later launches reuse installed requirements.
python scripts/setup_and_start.pyChartsmith is ready at http://localhost:3000/Open localhost:3000, keep the terminal open, and press Ctrl+C there when you want to stop.
Upload one PNG, JPEG, or WebP chart. Reconstruction, design directions, selection, rendering, inspection, and repair appear as user-facing stages. A full redesign can take tens of minutes.
No raw dataset or design brief required.
Read intermediate decisions while the workflow runs.
Chartsmith is not intended for high-stakes, pixel-perfect data extraction.
The launcher checks each requirement before it opens the studio.
Install Python 3.11 or newer, enable “Add Python to PATH,” open a new terminal, and run python --version.
Install Node.js 22.13 or newer, open a new terminal, and runnode --version and npm --version.
Run the exact login command printed by Chartsmith, finish the browser flow, and rerun setup. A global install uses codex login. Confirm it with codex login status.
Stop the other local Chartsmith or development process using that port, then retry.
Check the connection and run the same command again. Completed requirements are cached.
Run python scripts/setup_and_start.py --check. It checks the workflow, ten journeys, assets, test images, and reference index.