What this directory helps you decide
Awesome Jev is an independent community field guide to TypeSafe's Jev. Start here when you need to find a client, integration or evaluation report, then follow the listing to its own evidence. The repository explicitly disclaims affiliation with and endorsement by TypeSafe AI.
The guide distinguishes bounded model judgments from application policy. Its examples introduce Choice, Noul and Score, while application code sets thresholds and performs actions. The displayed support-ticket response is a saved jev-1.13.0 example. It is not evidence of a live request made by this review.
That distinction gives readers a useful next step: choose a candidate integration, inspect what it sends off-device, and test its answer-to-action mapping on representative labelled cases. Browsing an entry does not complete those checks.
What our dated sandbox run establishes
Our run on October 4, 2026 used commit af429b4. In our sandbox, installation took a recorded 9 seconds, the build step took 2 seconds and the test step took 5 seconds. The harness stores each step's outcome, exit code, elapsed time and timeout status separately. All three steps returned zero without a recorded timeout.
The attached stored parser summary reports 2 passed tests and no failures, skips or errors. The raw tail repeats the same pytest footer under a summary label; it does not describe another set of passing tests. No new execution or complete-log reconciliation was performed for this correction, and the parser summary does not prove exhaustive unique coverage.
The zero exit codes establish the recorded command outcomes. They do not establish a deployed website, a successful Jev API call, the correctness of a linked model, or the security of a community integration. Our evidence contains no workload evaluation of those projects.
The recorded pip-audit total is 0 advisories, with unknown scan coverage. Treat that as an advisory result with a coverage gap, not a clean bill of health.
MrKeyoor's testing methodology explains the distinction between command outcomes, parsed counts and workload validation. That page is provenance context; its current procedure does not establish which methodology version governed this older run.
What the contribution rules require
The contribution guide asks for a canonical public URL, a license for code, an explanation of data leaving the machine and one material limitation. Benchmark submissions must include their method, model version, task data, results or sufficient aggregates, and limitations.
These requirements help a reader identify what to investigate. They do not mean maintainers reproduced every submitted result. The guide explicitly says that being listed does not imply endorsement or a security review.
For a first resource pull request, contributors can edit README.md alone. Maintainers generate the derived directory, pages and cards before merging, and CI checks that the committed artifacts match the README. This provides a documented route for correcting a stale description without manually editing generated pages.
When another resource answers the question better
JevBench measures typed decision accuracy, probability calibration, latency and cost. Use it alongside Awesome Jev when the question changes from finding projects to examining model comparisons. Its own limitations include finite task coverage, sealed inputs that prevent complete independent reruns, opaque hosted endpoint versions and modeled costs. Its results are external evidence, not results from our October 4 sandbox run.
Awesome indexes topic lists across programming languages, development tools and other subjects. Choose it for broader resource discovery. It is an alternative browsing starting point, not a Jev evaluation service or a replacement for checking an integration's source.
How to read the ratings
These are editorial ratings of the directory's documented workflow, not measurements of model quality or adoption. The scale is 1 for absent or obstructive support, 2 for substantial gaps, 3 for a usable basic path, 4 for a documented path with maintenance checks, and 5 for a direct path with explicit instructions and limits. Each dimension uses that scale for its stated scope.
Setup ease concerns reading and contributing to the directory. Documentation concerns the supplied explanations and contribution instructions. Community concerns the documented route for participation and correction, not community size or responsiveness. Maturity concerns README-derived publishing and maintenance checks; the evidence does not establish a long operational history.

