Can I run the examples without a TypeSafe key?

Every level from 0 onward calls the real Jev API, so request access, generate a key, and set TYPESAFE_API_KEY; the written levels alone do not execute decisions.

Do I need a paid LLM key for the writing half?

Any LiteLLM provider string works for the write half in Levels 6 and 10, including free keys, local Ollama, or Ollama Cloud proxied models.

How is the capstone different from earlier levels?

Levels 0 through 9 isolate one idea in single scripts, while Level 10 is an importable capstone package with client, pipeline, CLI, unit tests, and evals.

What is the difference between its tests and evals?

Unit tests are fast, free, and mocked, while evals are real API calls against a golden dataset tracked over time rather than pass or fail gates.

How do I set up and run the first example?

Install uv, run uv sync, copy .env.example to .env and fill in your keys, then run uv run examples/00_hello_world.py. A probability and a millisecond timing mean you are set up.

What is the difference between the tests and evals directories?

tests/ holds fast, free, mocked unit tests for the capstone, run with uv run pytest. evals/ runs quality checks against a golden dataset using real API calls and tracks them over time rather than passing or failing.

Which provider is used for the write half of the course?

Levels 6 and 10 use a regular LLM for drafting text, chosen via CHAT_MODEL through LiteLLM so any provider works. Jev only decides; it never writes prose.

CommunityJev Benchmarks & Evaluations

Jev Crash Course — nadyth

Eleven-level Jev crash course from zero with runnable examples, LiteLLM write-half, and a ticket-triage capstone with tests and evals.

See highlights

Open github.com

Stars: 0
Language: Python
License: MIT
Published by: Community
Author: nadyth
Last verified: Oct 2026

nadyth/jev-crash-course

User repository on GitHub

View on GitHub

An 11-level crash course on Jev, TypeSafe AI's System One decision model — runnable examples against the real API, plus a capstone project with unit tests and evals. Works with any LLM provider via LiteLLM.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

This is one of several Jev listings for the same project. The canonical entry is Jev Crash Course — nadeemcite.

About this project

Traditional software can only handle the branches you hardcoded. Agentic systems hand every step to an LLM — flexible, but slow, expensive, and hard to predict. Smart software is the middle path: a fast, cheap decision model at each branch point, with the LLM reserved for the one step that actually needs to write something. That third column is what Jev is for.

Quoted from the project's README. Read the full README.

Tagscommunity github

Works forJev for Evaluation

Used inJev for Software & DevTools

Languages: Python

Highlights

Teaches state plus typed noul, choice, and score questions with calibrated probabilities.
Uses uv, Python 3.12+, and LiteLLM so any chat provider works for the write half.
Pairs levels 0, 5 through 7, and 10 with runnable scripts including timing comparisons.
Ends in a ticket-triage capstone with importable package, mocked tests, and live evals.

Watch out

Runnable course needing an early-access key from Level 0 on; written levels explain concepts but examples call the real API. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

jev-crash-course

Eleven-level Jev course from hello-world to a ticket-triage capstone, pairing short lessons with runnable scripts against the live API and using LiteLLM so the writing half works with any provider.

Jev Benchmarks & Evaluations#community#github

Community

nadeemcite/jev-crash-coursePython

jev-sim

Wire-compatible Jev re-implementation reading typed decisions from an LLM's next-token logits, plus a reproducible JevBench-based comparison of that approach against Jev on identical items.

Jev Benchmarks & Evaluations#community#github

Community

dashbi1/jev-simPython

apple-silicon-systemone-benchmark

Native Apple Silicon harness comparing local decision models through one decision interface, running models serially and emitting an offline dashboard with latency and accuracy.

Jev Benchmarks & Evaluations#community#github

Community

Giamme/apple-silicon-systemone-benchmarkPython

SREGym

338

SREGym stands up live Kubernetes clusters so SRE agents can be evaluated against 90 fault scenarios, including OS-level faults and metastable and concurrent failures, in containers that cannot see the grading logic.

Jev Benchmarks & Evaluations#community#jevbest

Community

SREGym/SREGymPython

tasksource

217

Standardizes hundreds of public datasets into interchangeable multiple-choice, classification, and token-classification templates, and deterministically recasts them into typed choice, score, and noul decision requests.

Jev Benchmarks & Evaluations#community#jevbest

Community

sileod/tasksourcePython

hev reranker

A BEIR reranking recipe that asks up to 30 parallel Noul relevance questions in one Jev state, with RESULTS.md against cross-encoders and hosted rerankers.

Jev Benchmarks & Evaluations#community#evaluation#python

Community

hev/rerankerPython

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai