jev-review-action
Configurable GitHub Action where Jev reviews catalog submissions or classifies PRs under a JSON policy and updates one templated comment.
Overview
Jev Review Action reviews submissions and classifies pull requests with TypeSafe Jev. You define criteria and categories in a JSON policy; Jev returns typed judgments; code applies the policy and updates a single PR comment from a fixed template. Only Jev is used: no text-generation model, autonomous agent or hosted bot. Jev can be reached through TypeSafe, Vercel AI Gateway or Cloudflare Workers AI with an ordered fallback chain.
Catalog mode reviews 1-10 submitted JSON entries using each entry's public repository README and source files at an immutable commit, suited to awesome lists and directories; its first consumer is the Awesome Jev list. Pull-request mode uses the PR title, body and changed-file patches for scope checks and change classification. Policies are not hardcoded to the Jev ecosystem. The workflow runs via pull_request_target using only trusted base-branch code, verifies the checkout matches the base SHA, reads submitted data through the API, rejects symlinks and submodules, and never executes PR content.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which category does this submission or PR belong to? | — |
| Yes / no | Does this submission meet this policy criterion? | yes · no |
How it works
- Add a JSON policy and workflow to the default branch
- On a PR, gather submission evidence or PR patches via the API
- Ask Jev the policy's typed questions
- Apply the policy in code and update a templated comment
Strengths
- No generative model or hosted bot
- Secure handling of fork submissions
- Provider fallback chain
- Topic-agnostic policies
Caveats
- Policy design determines quality
- Requires a Jev provider key as a secret
Getting started
Copy an example policy to .github/jev-review.json, set TYPESAFE_API_KEY as a secret, and add the documented workflow using fatwang2/jev-review-action.At a glance
- Type
- Project
- Status
- beta
- Language
- TypeScript
- Runtime
- GitHub Action
- Has code
- yes
- Has tests
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
A configurable GitHub Action that reviews catalog submissions or classifies PRs and updates a templated comment.
logicrw/awesome-jev-projects
Frequently asked questions
What does Jev decide?
Policy checks and categories for catalog submissions or pull requests.
Does it run PR code?
No. It reads data via the API and never executes PR content.
Which providers are supported?
TypeSafe, Vercel AI Gateway and Cloudflare Workers AI, in fallback order.
Related projects
Local software factory running YAML pipelines of coding agents and shell steps, with a typesafe step for typed Jev judgments.
ProjectbetaPythonGlue that turns one Linux machine into a team of coding agents, with Jev making typed calls like who builds a task and whether it may merge.
ProjectbetaTypeScriptLabels GitHub pull requests with Jev via OpenRouter: type, area, platform, attention and a size based on conceptual scope.
ProjectbetaPythonSmall Python framework and GitHub Action for Jev-powered workflows, with issue triage and PR screening built as examples.
ProjectbetaPythonAutomated PR review: an LLM agent finds issues, a verifier refutes them, and Jev decides how to triage, route and label.
ProjectproductionTypeScriptRust code-review gate asking Jev small typed questions about functions and tests, failing CI only on rules measured 80% right.
CLIproductionRust