Skip to content

README.md

Latest commit

 

History

History
53 lines (36 loc) · 1.18 KB

README.md

File metadata and controls

53 lines (36 loc) · 1.18 KB

Benchmark Task Studio

A frontend-only React application that demonstrates how real-world coding benchmark tasks can be authored, reviewed, and validated.

This project simulates the workflow used in AI evaluation and developer benchmarking platforms, using mocked task data (no backend or authentication).

What this project demonstrates

  • How benchmark tasks are structured with:
    • Problem description
    • Bug explanation
    • Fix approach
    • Test intent and guarantees
  • Clear separation between buggy and fixed implementations
  • How reviewers reason about correctness, performance, and regressions
  • Clean internal-tool style UI focused on clarity over visuals

Tech Stack

  • React + TypeScript
  • Vite
  • React Router
  • Plain CSS (no UI libraries)

Features

  • Task list with search and filters
  • Task detail page with:
    • Problem context
    • Review panel (bug summary, root cause, fix, test guarantees)
    • Side-by-side buggy vs fixed code comparison
  • Fully frontend (mocked data)

Routes

  • / — Task list (search + filters)
  • /tasks/:id — Task detail with review panel and code comparison

Running locally

npm install
npm run dev