Jevify skill

The stated intent is report first: it marks which existing LLM calls could move, then converts them. Run it on a clean branch and read the marked list before letting it touch a repository you actually depend on.

Per the author, the scan runs without a Jev key when you only want a report. That makes it a cheap way to inventory candidate calls before you commit to a provider or change any billing wiring.

It is an independent skill, and the author states plainly that they are not affiliated with TypeSafe. It is a convenience wrapper rather than a vendor tool, which means you own the review of every conversion it proposes.

Thinly. The author reports testing on a single popular repository plus a demo, and the thread carries almost no discussion of failures. Treat it as unproven on your stack, especially where the original call produced prose rather than a structured label.

The author tried Jev when it came out, found it very fast, and noticed many LLM calls in their agents were already structured data. They moved those over and wrapped the end-to-end process in a skill you run with /jevify.

TypeSafe, OpenRouter or Vercel, per the author. The scan marks which calls could move to Jev first, then converts the ones you keep.

Very little: 2 points and essentially no discussion thread, so there is little independent review beyond the author's own demo and single test repository.