Awesome Jev
'Awesome Jev' is what a growing shelf of community lists call their directories of things built on Jev — TypeSafe AI's System One decision model. This is our own source-backed tour: real projects worth opening, the community lists to go deeper, and the tools on this site you can try right now.
Jev returns a typed decision — a choice, a score, or a calibrated yes/no (noul) — in one call, so it slots into anything that needs a fast, bounded decision instead of generated prose. That's why the ecosystem filled up quickly: browser agents that pick the next click, routers, judges, guardrails, local runtimes and benchmarks. Below is a curated map, with the independent community lists linked at the end.
Browser & computer-use agents
The breakout category: let a cheap, fast Jev call pick the next action or element, and reserve an LLM for the parts that need reading or typing.
- fastbrowse — Jev picks each browser action from what's on the page; an LLM reads and plans
- jev-ultrafast — browser-use's ultrafast variant: Jev decides each click, the LLM is called only to type
- jev-use — voice-driven macOS computer use: speak a goal, Jev picks the next on-screen action from the accessibility tree (no screenshots)
- Sedum — goal-mode browsing and authored test steps: Jev picks the element and judges whether a claim holds, Playwright acts
Try the ones running on this site
Everything here is made with Jev over the same hosted endpoint you get with a key — open one, then build your own.
Run it yourself & measure it
- Ollaya — the 'Ollama for decision models': run open Jev-like models (Laya, winnow) locally — see /jev-ollama
- jev-mcp — give any MCP client a typed decide tool with one npx command — see /jev-mcp
- jev-bedrock — deploy a Jev decision action group for an Amazon Bedrock agent — see /jev-aws-bedrock
- JevBench — independent leaderboards for Jev-class decision models — see /jevbench
The community lists (not ours)
Several developers maintain their own 'awesome-jev' directories — independent, community-built, not by us or TypeSafe. They're the best way to go wide:
- cobanov/awesome-jev — a curated, source-backed list of projects built with Jev
- awesomejev.com — a directory site cataloguing 1,000+ Jev projects, SDKs, demos and benchmarks
- kraayenjon/awesome-jev — use cases, projects, SDKs and learning resources
- AbdelStark/awesome-typesafe-jev — a source-backed field guide with SDKs, demos and independent evaluations
FAQ
What is 'awesome-jev'?
It's the name several community developers use for their curated directories of projects built on Jev, TypeSafe AI's System One decision model — browser agents, SDKs, integrations, demos and benchmarks. This page is our own source-backed tour plus links to those community lists; we don't own the 'awesome-jev' name.
What kinds of things are built with Jev?
Mostly fast, bounded decisions: browser and computer-use agents that pick the next action, model routers, LLM-as-a-judge evaluators, injection and fraud guardrails, semantic code search, and local runtimes. Anything that needs a typed answer instead of generated text is a fit.
Can I try some without installing anything?
Yes — the tools and demos on this site (PR Judge, Model Router, jevgrep and more) run on the hosted Jev endpoint right in the browser. Open any of them, then grab a jv_live_ key to build your own.
Where's the biggest list?
awesomejev.com catalogues over a thousand projects, and several GitHub 'awesome-jev' repos (cobanov, kraayenjon, AbdelStark and others) keep their own source-backed directories. All are community-maintained, not official.
See also: Made with Jev · All tools · jevgrep · JevBench · Playground
Build something awesome with Jev
Try a real decision in the browser free — no signup — then grab a jv_live_ key and ship your own.