Jev use cases
The patterns builders actually search for: moderation, routing, triage, RAG verify, and agent gates.
Moderation & trust/safety
Discord bots, comment guard demos, indie media trust gates, and scam filters use parallel Noul and Choice questions to escalate or block content with explicit confidence.
Model & tool routing
jev-router, tiershift, hono-jev-router, and gateway routers pick models or tools from structured catalogs, often sub-200ms decisions per turn.
Support triage & chargeback gates
LaneBreak, CartShield, and typesafe-triage-guard demonstrate ticket priority, refund disposition, and deploy-risk gates without generative summaries.
RAG verify & rerank
llama-index-jev, jev-rerank, and citation-verifier projects use Score and Noul to rerank or verify claims cheaper than LLM-as-judge.
Agent auto-mode & compaction
fast-jev-compaction and pi-jev-auto-mode score tool results and approvals so agents keep verbatim context or auto-approve safe tool calls.
Browser and desktop automation
Computer-use demos treat the next click or keypress as a Choice over a finite action catalog. Jev scores candidate nodes quickly enough for interactive loops. Pair with vision or DOM snapshots as your state payload.
Games and simulations
Arcade demos prove System One latency in public. They are useful sandboxes before you wire the same primitives into billing or safety code paths.
How to explore further
Use category filters on Explore to narrow listings by pattern. Favor repos with demos or recent commits when you are evaluating fit. Submit your own project if you ship something missing here.
Know someone still labeling with chat?
Send them this guide. Low spam, high signal.
Drop it in Slack. Tag TypeSafe if it saves someone a sprint.
More context on Jev Directory and the llms.txt map for AI crawlers.