What it does
It targets the failure modes syntax rules miss — patchwork structure, swallowed errors, placeholders, duplication, over-engineering — and so works on any language.
Agent tooling
A linter that sends each file an agent touched to Jev and asks one question: is this slop?
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
It targets the failure modes syntax rules miss — patchwork structure, swallowed errors, placeholders, duplication, over-engineering — and so works on any language.
A developer can use this project to check if an AI writing assistant is making a mess. The tool sends newly written files to an outside service called Jev. Jev looks for common mistakes like copied text or ignored errors, then prints a report.
Connecting to Jev requires a paid access key. Right now, the tool only prints advice and cannot fix the mistakes itself. It also reviews files one by one, which takes extra time if you have thousands of documents.