User repository on GitHub
Uses typesafe jev, zero sync to pull and sync large repositories for issue triage
- Forks
- 0
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Tools & CLIs
Typeful Triage
A multiplayer triage dashboard for public GitHub repos: issues and PRs sync into Postgres, Rocicorp Zero replicates them live, and Jev answers typed questions about each one, including what a maintainer should do next, with human corrections kept.
- Stars
- 4
- Language
- TypeScript
- License
- MIT
- Reading time
- 4 min
- Published by
- Community
- Author
- cephalization
- Last verified
- Sep 2026
About this project
A multiplayer triage dashboard for public GitHub repositories. Issues and pull requests sync into Postgres, Rocicorp Zero replicates them live to every browser, and a TypeSafe System One model answers a fixed set of typed questions about each one: what kind of issue it is, how severe, how urgent, whether it duplicates another, and above all what a maintainer should do next. People confirm or correct those answers, claim issues, and mark them done; every correction is kept and shown to the model on later runs.
Quoted from the project's README. Read the full README.
Used inJev for Data & Analytics
Highlights
- Jev classifies kind, severity, urgency, duplication, and the suggested next step; corrections are kept and shown to the model on later runs.
- Nothing is written back to GitHub; the app is a shared view over a repository, not a bot.
- An Unsure view lists issues Jev was not confident about or where a person disagreed, least confident first.
- Guided reviews have Jev classify every changed file by role, risk, reading care, and whether the change starts there before a model writes the walkthrough.
- Vite+ pins and fetches Node and the package manager on first use; Postgres, the reviewer cell, and Arize Phoenix run from one compose file.
Quickstart
bash
git clone https://github.com/cephalization/jev-triage
cd jev-triage
cp .env.example .env
vp install
vp run devWatch out
MIT-licensed; needs Vite+, a Docker-compatible CLI (Docker Desktop, OrbStack, or Podman), and a TypeSafe API key, with an optional GitHub token for pull request sync and higher rate limits.
Questions about this entry
Short answers drawn from the linked project or page.
Nothing is written back to GitHub. Issues and pull requests sync into Postgres and replicate to browsers through Rocicorp Zero as a shared view, and replies, approvals and merges still happen on GitHub. Marking an issue done only clears it from your queue, and handled work returns after a newer GitHub update.
Access is invite only. Put your GitHub login in ADMIN_GITHUB_LOGINS and admins listed there can invite other accounts with roles from System, People. Without a GitHub account, vp run dev:emulate starts a GitHub emulator and makes admin an admin, though repository sync still hits real github.com.
An Unsure view lists issues the model was not confident about, or where a person disagreed with it, least confident first. Every correction is kept and shown to the model on later runs, and the System view holds model calibration against that human feedback so drift is visible rather than silent.
No second pass. Recommendations reuse existing classifications and GitHub metadata. Every open issue and pull request is fetched and classified as fast as GitHub and TypeSafe allow, which the README puts at well under a minute and a few cents for a thousand issues, with no sync or classification knobs beyond pausing one repository.
Vite+ (installed with the curl script), a Docker-compatible CLI providing docker compose, and a TypeSafe API key, with a GitHub token optional for PR sync. Then clone, copy .env.example to .env with your keys, vp install and vp run dev, and open http://localhost:5173 and add owner/name in the repository switcher.
Issues and pull requests sync into Postgres, and Rocicorp Zero replicates them live to every browser; apps/api is a Hono server holding the Zero endpoints, GitHub sync and the classifier worker, apps/reviewer is a celld-guided review environment running the pi SDK, and packages/triage holds the TypeSafe question builders and the pure decide() policy.
A TypeSafe System One model answers a fixed set of typed questions per item: what kind of issue it is, how severe, how urgent, whether it duplicates another, and what a maintainer should do next. People confirm or correct those answers, claim issues and mark them done, and every correction is kept and shown to the model on later runs.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Decision outputs in Grep
Hacker News: jevmem keeps Claude Code project memory by asking Jev which turns are worth saving.
avinashjetwani61 pointsJevmem – automatic project memory for Claude Code, built on JevRead the thread on Hacker NewsX: A usage router for Grok Bot
X: The compaction plugin, in Chinese
More Jev resources like this
A TypeScript layer beside an LLM: JevSift selects approaches, JevLoop audits progress with typed actions, and JevLong watches sessions for stalls without steering them.
Jev Tools & CLIs#community#typescript#coding-agent
Community
Alex314618-create/JevRevTypeScript
Alex314618-create/JevRevTypeScriptAn open-source, commercially hosted generative-engine-optimization platform that uses Jev judgments within a broader application to track brand mentions and placement in AI answers across ChatGPT, Claude, Gemini, and Perplexity.
Jev Tools & CLIs#community#typescript#postgres
Community
usenotra/notraTypeScript
usenotra/notraTypeScriptA 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 router that turns models, subagents, skills, MCP tools, CLIs, and plugins into one candidate set Jev answers with a single typed Choice, while the router code owns availability, permissions, and confirmation.
Jev Tools & CLIs#community#typescript#routing
Community
BillionsBobby/JevRouterTypeScript
BillionsBobby/JevRouterTypeScriptA 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
Community
devagrawal09/stanley-codeTypeScript
devagrawal09/stanley-codeTypeScriptA research-stage proposal-review contract in TypeScript: an LLM proposes one action, Jev answers four pinned questions, and code produces evidence the host can consider.
Jev Tools & CLIs#community#typescript#agents
Community
TypeSafeAI/jev-harnessTypeScript
TypeSafeAI/jev-harnessTypeScriptFrom 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.