JevMade Sign in
← Back to experiments

Benchmarks & research

jev-shadcn-lint-eval

Source screenshot of jev-shadcn-lint-eval
SOURCE SCREENSHOTFull screenshot ↗

What it does

A test of whether Jev can tell genuine shadcn-ui lint problems from false alarms.

How you can use it

Use this project's questions to check the warnings your website team receives. Do the warnings say what needs changing, or only that something is wrong? A developer can ask AI to judge them. Include examples that should not trigger a warning, too: the AI may mistake allowed designs for errors.

Maker-reported (not independently measured by JevMade): weaker on clean code and flagged 16 of 52 clean cases. Overall it answered 82% · The run made 246 requests with no errors and used 249,384 input tokens (Jev

Primitives
score
Platform
JavaScript
Added
Project created
GitHub stars
0 (snapshot, not live)