Every AI Finder
All benchmarks
Protocol ready · controlled run pending

Cursor vs Copilot vs Replit: Coding Test

Build the same small TypeScript expense tracker and run an identical test suite.

Common prompt

Build a TypeScript expense tracker with add, edit, delete, category filtering, monthly totals, local persistence, keyboard-accessible forms, and tests for calculation and invalid input. Do not add a backend or external UI library.

Controlled conditions

  • Same empty repository and dependency lock file.
  • No copied starter code beyond the test fixture.
  • Stop the timer when all required tests pass.

Published criteria

Task completion

35%

Builds every required behavior without skipped requirements.

Correctness

30%

Passes the same tests and avoids runtime errors.

Maintainability

20%

Reviews structure, naming, unnecessary dependencies, and explanation.

Time to working result

15%

Measures prompt-to-running-result time and manual intervention.

Tool records

A controlled test has not been run yet. Raw outputs and scores are therefore hidden and do not affect tool ratings.

Cursor

Plan: Controlled test plan to be recorded

Protocol ready · controlled run pending

Tool details

GitHub Copilot

Plan: Controlled test plan to be recorded

Protocol ready · controlled run pending

Tool details

Replit

Plan: Controlled test plan to be recorded

Protocol ready · controlled run pending

Tool details