SuperAPI AI Directory

research-ui-prototyping

Frontend & websites

View repository

Updated 10-16 03:33

About this skill

Research UI Prototyping

Research UIs are working tools, not landing pages. Favor dense, inspectable, task-focused interfaces that expose evidence and uncertainty.

Read First

  • references/repository-contract.md
  • references/output-contracts.md

When To Use

  • dashboard for experiment metrics
  • source or paper browser
  • annotation or coding interface
  • SOTA matrix explorer
  • reproducibility report viewer
  • demo for a method or dataset

Design Rules

  • Start with the core workflow, not a hero section.
  • Show identifiers, provenance, filters, and export controls.
  • Make uncertainty and missing evidence visible.
  • Prefer tables, facets, timelines, charts, and diff views over marketing cards.
  • Keep visual style restrained and domain-appropriate.
  • Use the existing frontend-design skill when available for polish and UI craft.
  • Verify with screenshots and interaction checks before claiming completion.

Data Rules

  • Do not hardcode research results into the UI unless it is a static report.
  • Keep data loading paths documented.
  • Export tables/figures back to outputs/ when they support the paper.

Output

Record UI purpose, run command, and data dependencies in docs/agent/ or docs/reproducibility/commands.md.