Build interactive data stories with AI
Develop a sourced visual argument through a brief, storyboard and interactive implementation, with staged review of evidence, mobile behaviour and the final result.
Build interactive data stories with AI around a question that the evidence can support. The pudding-data-story skill develops the thesis, narrative sequence and visual argument before production code, keeping a source record for factual claims.
Its goal is a reader-facing story rather than a dashboard with scroll effects. Overall design, detailed storyboards and the evidence package have separate approval stages. New standalone projects follow a Svelte 5 baseline, while an existing stack can retain a documented exception. Motion should explain a change and preserve the explanation under reduced motion. The guide borrows editorial principles from reference stories, not their branding. Agree the publication plan early and keep private or unauthorised data out of a public repository.
What it does
- Define the question and central claim.
- Record sources and data transformations.
- Design the scene sequence in a storyboard.
- Make interaction support a real comparison.
- Review mobile and reduced-motion behaviour.
Install and use
- Install the skill in a compatible host and prepare authorised data, sources and an audience brief. New projects need the Svelte 5 build toolchain; check starter-code terms and remove protected branding.
- Ask: “Plan an interactive story from public library-usage data, explaining changes across the year. Review the question and evidence first.” State scope and publication constraints.
- Approve the design, manuscript and storyboard in stages, then inspect the preview. Verify numbers, controls, backscroll and mobile layouts. Publish only within the agreed project plan and permissions.
FAQ
Source
Made by zhiqiangzeng. Original skill: pudding-data-story, published on GitHub under the MIT licence.
View on GitHub