NetworkChuck Jev guide
NetworkChuck's MIT Jev guide: five chapters from Playground spam checks to a Bun model router, tested on jev-1.14.0.
Overview
jev-guide (github.com/theNetworkChuck/jev-guide, MIT) is NetworkChuck's companion to his YouTube video (https://www.youtube.com/watch?v=Ck7wB7pdebQ, published October 28, 2026). The README states it is not affiliated with TypeSafe. Five markdown chapters walk from Playground spam detection through terminal API calls, TypeSafe's official coding-agent skill, a Jev model router, and a survey of community projects. Examples were tested on jev-1.14.0.
Problem: Newcomers want a guided path from the TypeSafe Playground to their own API calls and routing patterns without reading only official docs.
Built for: Developers following NetworkChuck's October 2026 Jev video who want a five-chapter MIT repo with copy-paste steps, spam-check scripts, and a Bun model router.
First indexed on Jev Directory: 2026-10-27
Creator and team
- Name
- NetworkChuck
- Handle
- @theNetworkChuck
How Jev is used
- Role in the product flow
- Teaching path: noul spam gates, multi-question batches, choice routing to small, medium, or large chat models via OpenRouter
- Primitives
- ChoiceScoreNoul
- State in
- Plain-text state (for example a YouTube comment) plus JSON questions in Playground or POST /v1/systemone from curl and Bun scripts.
- Decision out
- Structured probabilities and picks (noul 0 to 1, choice label with probabilities, score levels) without generated prose.
- Chapter 1: paste spam comment state and run noul, choice, and score questions in console.typesafe.ai/playground
- Chapter 2: export TYPESAFE_API_KEY and run examples/spam-check.sh with six bundled questions
- Chapter 3: install TypeSafe's official agent skill and ask where Jev fits your repo
- Chapter 4: run examples/router/router.ts so Jev choice picks small, medium, or large OpenRouter models
- Chapter 5: browse projects from the video and open-source Jev-shaped models
This listing is a third-party video course, not TypeSafe documentation. For authoritative primitives and API contracts, use this site's learn guides at https://jev.aitools.fyi/learn/jev-typesafe and https://jev.aitools.fyi/learn/system-one alongside docs.typesafe.ai. The repo complements those pages with NetworkChuck's spam comment narrative, bundled questions.json, and a Bun router that thresholds Jev confidence (demo SURE_ENOUGH 0.6, not a TypeSafe constant). datawhalechina-jev-cookbook is a longer Chinese Jupyter course; disler-ten-levels-of-jev is a TypeScript lab; gargpratyush-jev-router is a different production routing story. Cite https://github.com/theNetworkChuck/jev-guide?ref=jev.aitools.fyi and https://www.youtube.com/watch?v=Ck7wB7pdebQ&ref=jev.aitools.fyi. Requires curl 7.76+, jq, Bun for chapter 4; optional OpenRouter key for full router answers.
Sourced performance claims
- Playground multi-question run on the sample spam comment: spam noul about 96%, contact about 92%, money about 12%, link about 94%.Source: github.com/theNetworkChuck/jev-guide guide/01-playground.md (jev-1.14.0 run, may vary)
- Guide cites TypeSafe pricing on jev-1.14.0 at $0.042 per million input tokens with output free; spam-check usage example uses 529 input tokens.Source: guide/02-first-api-call.md referencing docs.typesafe.ai/models
- Model router demo: Jev choice decisions in about 110 to 170 ms at about $0.0000159 per decision on sample requests.Source: guide/04-model-router.md (jev-1.14.0 run, may vary)
Features and stack
Features
- Five ordered chapters from Playground to community projects
- examples/spam-check.sh and questions.json from the video
- examples/router/router.ts with --decide-only mode
- Tested on jev-1.14.0 per README badge
- MIT license; explicit not affiliated with TypeSafe disclaimer
Stack
- TypeSafe Playground and api.typesafe.ai
- curl and jq
- Bun for router.ts
- Optional OpenRouter for downstream chat models
Pricing: Guide repo is MIT. Live Jev calls spend your TypeSafe account; chapter 4 optionally spends OpenRouter credit when not using --decide-only.
Links
FAQ
- Is this guide official TypeSafe material?
- No. The README states the guide is not affiliated with TypeSafe. TypeSafe and Jev are trademarks of TypeSafe AI. Use jev.aitools.fyi learn pages and docs.typesafe.ai for authoritative docs.
- Where should I start if I already read TypeSafe docs?
- Use chapter 1 if you want the Playground spam walkthrough from the video, or jump to chapter 2 for terminal API shape. Pair with /learn/jev-typesafe on this site for concepts the video assumes.
- Do the spam percentages match every run?
- The guide warns numbers move slightly between runs and Jev versions. Documented spam, contact, money, and link nouls are from real jev-1.14.0 runs in the repo, meant as about this much, not exact targets.
- What does the router cost?
- Chapter 4 cites about 110 to 170 ms and about $0.0000159 per routing decision on sample requests, plus separate OpenRouter spend when you run full answers. --decide-only skips downstream model cost.
Related learn guides
Original Jev guidance that pairs with this product pattern.
- Jev & TypeSafe
A primer for builders who need to know what Jev is, who it is for, and when TypeSafe docs should take over from this directory.
- System One model
The model family behind Jev: parallel typed questions, one forward pass, probabilities you can threshold.
Related products
Hand-picked neighbors with rich profiles or overlapping tags.