askjev

How we reviewed this: we built it in Docker and drove the server over MCP against a local stand-in — 35 tests passing, and the live call sequence matched the diagram in its README. We made no Jev calls.

What it does

askjev, by pZacca, exposes one tool. You pass a question in ordinary words and the server asks Jev what kind of question it is — yes/no, choice, or scale — and then asks it properly. That’s two calls, or three when a scale needs a rubric chosen.

The error handling is the nicest part. Ask “Which team owns it?” without options and you get a message naming the problem and the fix, word for word as documented; pass a single option and the schema rejects it before any call is made.

What to know

Two calls instead of one is the price of not making the caller decide — reasonable for an agent-facing tool, worth knowing if you’re cost-sensitive.

Nothing is redacted or truncated, and the local path writes nothing to disk. There is a hosted endpoint run by the author; using it means your key and your state pass through their domain, which the README states plainly.

Verdict

A small idea done carefully: one tool, a routing step, and errors that teach. Zero stars is an underestimate.

For a toolbox instead of a single tool, see Jev MCP.

See how it compares with other tools in Best Jev tools, tested hands-on.

Review updated Sep 20, 2026. Numbers quoted from the project are its author's own; we don't publish our own measurements of Jev.