Overview & result
mizchi shows jev-test-filter, built on Jev: Score every test against a git diff, then emit the runner arguments.
From the original post
jev-test-filter by @mizchi: Score every test against a git diff, then emit the runner arguments.
How Jev fits in the loop
- The tool reads a git diff and lists every test in the repository.
- One Jev request carries the diff as shared state and one question per test: how much can this change alter the test's outcome?
- It prints filter arguments for the runner you already use, such as vitest, jest, Playwright, cargo test or go test.
- With --exec it starts the runner itself, without a shell in between.
How to reproduce
- Review the source documentation and repository at https://github.com/mizchi/jev-test-filter.
- Configure environment with TypeSafe Jev / System One API keys.
- Run test prompts against the decision endpoints.
Why this build matters
Shows Jev in practical use for developer tools and apps.
Limitations
- The stdout form is shell-quoted and needs eval; the README recommends --exec instead.
- Rust tests are never discovered automatically, because listing them triggers a build.
- Requires Node 24+, git and a TypeSafe API key.