What it does
Findings can warn, request permission or block, with warnings as the default. Failed model checks are skipped. The project is in active development and declares MIT in package metadata, but has no license file.
Agent tooling
A coding-agent hook checks newly introduced names, using code for their shape and Jev for meaning and word order.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
Findings can warn, request permission or block, with warnings as the default. Failed model checks are skipped. The project is in active development and declares MIT in package metadata, but has no license file.
A developer can borrow these rules to watch an AI assistant while it writes code. Have the software check simple patterns like capital letters directly. Start by having the system show a warning about bad names instead of blocking work. This lets you see if the rules actually help.
Your developer can also connect the tool to an outside AI service called TypeSafe. This lets the software check what new words actually mean. You will need a unique text code from TypeSafe to authorize the link. The project lacks a legal license file, so confirm permission before copying anything.