GitHub · Coding and context · Triage and routing
devagrawal09/jev-review
Copying the shape of a review pipeline where the model only judges.
What it does
A staged code review, with the model answering only bounded questions: a Null risk matrix, then Choice and Score file profiles, evidence selection, mechanism classification, severity scoring, and a conditional Choice that routes which reviewer sees the result. The output lands in a quiet local dashboard.
The repository's own description: “A staged code-review workflow and local dashboard built with TypeSafe Jev.”
How it works
Orchestration stays in code and Jev answers the judgements; four npm scripts split the entry points (review:changes, review:codebase and the :save variants) plus a dashboard script, so a review is a command rather than a chat.
The repository, by the numbers
Read from the GitHub API on 2026-09-22. Stars and the last push move daily — quote them with the date, the way we do.
What we checked
- the tree: src, scripts, docs, tsconfig.json
- MIT LICENSE (Dev Agrawal)
- package.json scripts enumerate review:changes, review:codebase, their :save variants and dashboard
- the README publishes the exact judgement chain the reviewer follows
What we did not check. We have not pointed it at our own repository yet; the interesting part is the chain, and no accuracy figure is published.