Question 1

What is fuzzy-jev and what does it do?

Accepted Answer

A client for Jev and a rules engine on top of it: it asks typed questions about a piece of text, gets probabilities back, and combines them with fuzzy logic — AND, OR, NOT and hedges — into a support score per decision or a crisp amount. It ships as a CLI, a Rust library, a Python package and an Agent Skill.

Question 2

What are the three question types?

Accepted Answer

A Choice between named options, a Score on an ordered scale, and a Noul — the probability that something is true. Each question sees the same state and is answered on its own, so asking all of them in one call costs a few extra tokens and no extra round trip.

Question 3

What do the fuzzy rules add?

Accepted Answer

A rules file is a policy over the answers: each probability becomes the degree to which a condition holds, and the operators combine them into a support score per decision or a crisp output. The README is explicit that a support score belongs to the policy, not to Jev, and docs/rules.md works an example by hand.

Question 4

How do I install it?

Accepted Answer

The Python library is fuzzy-jev on PyPI (uv add fuzzy-jev or pip install fuzzy-jev), imported as jev, with wheels for Linux and macOS on CPython 3.10 and later. The jev command comes from cargo with the package name fuzzy-jev, since jev on crates.io is another project; it needs a Rust toolchain.

Question 5

What does a run cost?

Accepted Answer

The README's sample output shows 427 input and 73 output tokens against typesafe/jev-1.13-20260917 for $0.000018, using an OpenRouter key. The client reaches Jev through OpenRouter's decisions endpoint by default, and --url points the same questions at another endpoint.

Question 6

Can I use it as a library?

Accepted Answer

Yes: the Rust crate works natively and compiles for wasm32-unknown-unknown, and the Python guide is in docs/python.md. There is also an Agent Skill that teaches a coding agent when to ask Jev instead of judging by eye — jev add skill writes it into your project.

Question 7

How do I see the rules?

Accepted Answer

Pass --svg and the whole rule base is drawn as an SVG: each question's probabilities, each rule's operators and the outcome against its threshold. The README includes a weather example, docs/wear.svg, and docs/rules.md covers every table, operator, formula, output format and error.

Fuzzy Jev

Highlights

Quickstart

Questions about this entry

From the Jev guides

From the Jev AI model community