User repository on GitHub
DiffJury — TypeSafe Jev PR risk router + code review coach
- Forks
- 2
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Tools & CLIs
DiffJury
A Next.js app that fetches a public GitHub pull request's dossier and runs one Jev systemOne judgment, returning risk scores, noul probabilities, review depth, and an approve, request-changes, or block verdict.
- Stars
- 7
- Language
- TypeScript
- Reading time
- 3 min
- Published by
- Community
- Author
- raihankhan-rk
- Last verified
- Sep 2026
About this project
Jev decides if this PR is risky or not. Paste a public GitHub pull request URL and hit Analyze. One click fetches the PR dossier (title, body, diff, contributors) and runs a TypeSafe Jev systemOne judgment — risk bars, noul probabilities, and a verdict. No essays. No second button. Landing: one large URL field. Optional quiet sample chip. Analyze calls POST /api/analyze with the PR URL. Success opens a split view — dossier on the left, judgment on the right. Failure stays on the landing screen with a clear error.
Quoted from the project's README. Read the full README.
Works forJev for Code
Used inJev for Security & Privacy
Highlights
- One POST /api/analyze call fetches a public PR's title, body, diff, and contributors, then makes a single Jev systemOne call.
- The judgment returns nine typed answers, including risk, review_depth, needs_security, missing_tests, and a verdict.
- Large PRs are trimmed locally to fit Jev's ~32k/64k token limits; small PRs keep their full, unchanged context.
- Stores only an anonymous successful-analysis count; no PR URLs or user identities are persisted.
- Without a Railway /data volume the counter falls back to /tmp and reports persistence: temporary.
Quickstart
bash
npm install
cp .env.example .env.local
# set TYPESAFE_API_KEY=YOUR_API_KEY; GITHUB_TOKEN is optional
npm run devWatch out
No license file, so reuse terms are unclear; needs a server-side TypeSafe API key, only analyzes public repositories, and the README calls it a demo app.
Questions about this entry
Short answers drawn from the linked project or page.
The README calls it a demo app, and nothing here validates the scores. One systemOne call returns nine typed answers: risk, review_depth, needs_design, needs_security, merge_blocker, missing_tests, docs_debt, blast_radius, and a verdict of approve_with_nits, request_changes, or block. There is no labeled PR set, no reported precision, and no LLM fallback in v1.
Large PRs are trimmed locally to fit Jev's roughly 32k/64k token limits, while small PRs keep their full, unchanged context. The README does not describe the trimming strategy — head, tail, or diff-aware — and that detail matters a lot for a review judgment, so it is not documented here.
No. There is no GitHub OAuth and the fetch only works for public repositories; unauthenticated GitHub requests are the default. A GITHUB_TOKEN on the server only raises API rate limits. An invalid URL, a 404 or private repo, a rate limit, and a missing TYPESAFE_API_KEY each return an explicit error rather than a judgment.
The TypeSafe key is read only from process.env on the server through a POST /api/analyze route and never exposed to the browser. The app keeps one aggregate count of successful analyses, surfaced at /api/stats with a persistence field; no PR URLs and no user identities are stored. The repository has no license file, so reuse terms are unclear.
Nine typed answers: risk (Score), review_depth (Choice of skim, standard, deep), needs_design, needs_security and merge_blocker (Noul), missing_tests, docs_debt and blast_radius (Score), and a verdict of approve_with_nits, request_changes, or block.
npm install, copy .env.example to .env.local, put TYPESAFE_API_KEY (and optionally GITHUB_TOKEN) in it, then npm run dev and open http://localhost:3000; production locally is npm run build then npm run start.
GET /api/stats returns an anonymous successful-analysis count and a persistence field. It stores only that aggregate and no PR URLs or user identities, and without a Railway /data volume it falls back to /tmp and reports temporary.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
YouTube: Kubernetes admission control with a decision model, fail-closed.
Hacker News: SafeDep Gryph plus Jev checks for compromised coding-agent behavior.
2 pointsDetecting Compromised AI Coding Agents with Jev and GryphRead the thread on Hacker NewsX: Jev and Laya on security queries
X: Testing Jev for prompt injection
More Jev resources like this
Pi extension that supervises a coding agent with Jev judgments: it holds risky tool calls, checks writes against project Markdown rules, and feeds most issues back to the agent as a steer instead of interrupting you. The conscience is beta and off by default.
Jev Tools & CLIs#community#typescript#guardrails
Community
DevMortimer/pi-wardenTypeScript
DevMortimer/pi-wardenTypeScriptAgent-ergonomic CLI following the AXI conventions that gives coding agents Jev judgments for blocking risky tool calls, screening fetched content for prompt injection, triaging logs, flagging risky diffs, and ranking many items. Routine commands are decided locally at no cost.
Jev Tools & CLIs#community#typescript#cli
Community
shiftynick/jev-axiTypeScript
shiftynick/jev-axiTypeScriptA staged code-review workflow that uses Jev for bounded judgments over Git diffs or whole codebases, with a local dashboard. Policy and thresholds stay in code while Jev screens five risk areas.
Jev Tools & CLIs#community#typescript#code-review
Community
devagrawal09/jev-reviewTypeScript
devagrawal09/jev-reviewTypeScriptA guard that stands outside an LLM agent and uses TypeSafe Jev to score both incoming prompts and proposed actions, turning the score into a block, a question, or a pass; in the README's Gandalf test it rejected 17 of 21 published injection prompts before send.
Jev Tools & CLIs#community#github#security
Community
JamesANZ/JevPromptShieldTypeScript
JamesANZ/JevPromptShieldTypeScriptA local-first MCP server that gives Claude Code, Codex, Cursor, and OpenCode structured quality scores from Jev across correctness, complexity, tests, security, and other dimensions while the agent writes code.
Jev Tools & CLIs#community#typescript#mcp
Community
NiazMorshed2007/jev-reviewTypeScript
NiazMorshed2007/jev-reviewTypeScriptA Jev-first command-line coding agent: natural-language requests are routed by Jev to one of ten bounded review and triage workflows, and unsupported requests fall back to the Pi coding agent or a workflow Stanley stages for promotion.
Jev Tools & CLIs#community#typescript#cli
Communitydevagrawal09/stanley-codeTypeScript
devagrawal09/stanley-codeTypeScriptFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.