scrollytelling-data
You can say to the Agent “Run the Demo” to start the demo, or upload your own data.
- Run the Demo — builds a full story from the bundled sample workbook (
assets/Basketball_Demo_Data.xlsx, a multi-sheet basketball players + teams dataset). - Upload your own data — drop in a
.xlsx,.csv,.json, or a PDF with tables and the Agent builds the story from that instead.
What you get
A single self-contained .html file that opens in any browser with zero setup and tells your data’s story as you scroll:
- Narrative chapters generated from your data
- Interactive charts — map, bar, scatter, line, pie/donut
- Stat cards that count up, animated reveals, a progress bar, and chapter markers
How it works
The Agent walks through five phases — inspect the data, find the story, crunch the numbers, build the page, and deliver the file — narrating friendly progress along the way. You don’t need to configure anything; just ask for a data story (or say “Run the Demo”) and open the resulting file.
Under the hood it loads your data into SQLite and runs the group-bys, sorts, totals, and other aggregations there — so the numbers in your story are computed with a real query engine and come back accurate.
Files
SKILL.md— the full workflow the Agent followsassets/Basketball_Demo_Data.xlsx— the bundled demo datasetscripts/build_story.py— reference implementation of the page generatorreferences/— chart patterns and demo-mode notes