HOW TO READ THE LABELS

Methodology

How the index is organised: four evidence labels, four categories, and one destination hierarchy per entry.

Evidence labels

official

Published by TypeSafe or by the project's own owner: launch posts, docs, SDKs, evals, status.

source-backed

A public repository, documentation page or live artifact backs the entry.

author demo

A result shown by its author: a post, video or live demo. Figures belong to that author's setup.

commentary

Roundups, analysis, explainers and press.

Figures

Numbers in an entry's notes are quoted from the source with the context it gives: task, sample size, single call or end-to-end. Who measured is visible from the label. Finance, security and autonomy demos are marked high-risk.

X reach is a dated snapshot of the public view counter, not a unique-visitor count. The By reach sort places entries without a readable counter after measured entries.

Categories

• Official Jev: TypeSafe's model, docs, console and skills.
• Integrations: plugins, SDK wrappers, harness patterns and tools that call the official API.
• Demos: results shown by individual builders.
• Replicas: NanoJev, SemIf, local Qwen classifiers and other Jev-like models built outside TypeSafe.

Destinations

Each card opens the project itself: live service or demo first, repository second, documentation third. The original X post sits behind a separate X button. Entries known only from a post are marked X post only.

Content

Summaries and notes are written for Jevland. Entries are added by hand in batches; see the changelog.

Media

Some entries show one reduced-size still from the original post or article, next to the attributed source. Originals stay with their authors; an image is replaced or removed on the owner's request.

Last verified

Each entry shows the date its links were last checked.