What it is
Local PR review stack where Jev prioritizes human attention (P0/P1/P2) and an LLM explains behavior — without posting to GitHub.
What it does: Analyzes change units, ranks files by attention priority, and surfaces reports via CLI / skill / extension.
Why it belongs in the showcase: Complements jev-review with a local, non-posting attention-first workflow.
Caveats: Code leaves the machine to providers; priorities mean attention, not correctness; unit coverage is capped.
Evidence
Public MIT README; local CLI + agent skill + GitHub extension; priorities mean attention not correctness.
Caveats
- Sends changed code context to TypeSafe and OpenAI
- Does not post review comments to GitHub by design
- Distinct from devagrawal09/jev-review
- Coverage capped (default first 12 change units)
Unofficial case page. Stars / list position observed for heat; do not treat heat as a quality SLA. Review the upstream README before you depend on it.