The awesome list that treats curation as a warning label

What it does

awesome-jev is a curated directory of public projects, integrations, and discussions built specifically around Jev, TypeSafe AI’s System One model. It catalogs real-world applications where Jev acts as a software decision layer—taking unstructured state and a typed question, then returning a typed answer such as a boolean, choice, or score with an attached confidence. The repository aggregates entries across thirteen active categories, from model routing and agent guardrails to finance and content moderation, plus a catch-all bucket for credible public practice signals like X threads and Reddit discussions.

The interesting bit

Most awesome lists are passive link dumps; this one reads like a field guide with a trust-but-verify manifesto. The maintainer explicitly warns that inclusion satisfies only narrow rules—public, citable, genuinely uses Jev—and does not imply code quality, security, or even that the project runs at all. Same-day bulk submissions sharing a single scaffold are flagged as “unproven” despite meeting every formal criterion, a level of curatorial skepticism rarely seen in the awesome-list genre.

Key highlights

• Strict inclusion gate: entries must demonstrate a concrete typed-decision loop (typed question → typed answer with confidence → accept/reject/escalate), not merely resemble a classifier or router.
• Over 250 entries across 13 active categories, with a dedicated “Related Practices / Discussions” section capturing credible signals that lack standalone repositories.
• Explicitly tracks bulk-submission risk, warning that scaffolded repos with thin commit histories may ship “considerably more prose than code.”
• README is auto-generated by a Python build script from per-category files, keeping the aggregate homepage current without manual duplication.
• Maintains an open “Scientific Pipelines” category that is currently empty and still being seeded.

Caveats

• Inclusion is purely rule-based; the list does not verify that code compiles, tests pass, numbers reproduce, or licenses permit reuse.
• Several categories are sparsely populated (Compliance & Legal has one entry; Scientific Pipelines has none).
• The repository warns that some entries may be mostly prompt documents rather than runnable code, and that a missing license limits reuse.

Verdict

Worth bookmarking if you are evaluating Jev or building typed-decision pipelines and need to see which patterns have been tried in public. Skip it if you are looking for a vetted software marketplace or if typed AI decisions are not on your roadmap.