User repository on GitHub
Java client and CLI for the TypeSafe AI API. Support local models like CLEF and LAYA via ONNX Runtime
- Forks
- 1
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Repos & SDKs
TypeSafe Java Client
A Java 21 client and CLI for the TypeSafe API: send state with Noul, Choice, and Score questions, get typed answers, and optionally run local Clef and Laya models via ONNX Runtime with no API key.
- Stars
- 0
- Language
- Java
- License
- MIT
- Reading time
- 2 min
- Published by
- Community
- Author
- dfa1
- Last verified
- Oct 2026
- Languages
- Java
- Capabilities
- Open sourceRequires API key
Highlights
- Send a state plus Noul, Choice, and Score questions and read back typed answers, with the token read from ~/.typesafe.apikey or an environment variable.
- A typed-record mapping module lets a Java record annotated with @Noul decode a typed result, skipping maps and casts.
- The CLI ships as a self-contained uber-jar whose --min thresholds make it exit 1 on failure, so it doubles as a CI pass/fail gate.
- Published on Maven Central as io.github.dfa1.typesafe-java, with core, client mapping, and cli modules.
- Local models run through ONNX Runtime; an Apple M5 table compares Laya fp32 and others against jev-1.13.0 on 104 cached requests.
Quickstart
bash
java -jar typesafe-java-cli-*-all.jar \
--state "My card was charged twice." \
--noul "urgent=Is this urgent?" --min "urgent=0.5" || echo "not urgent enough"Watch out
Copy reviewed against the repo README on 2026-10-15; the local-model table was truncated in the fetched excerpt, so the full agreement figures were not reviewed. Local-model agreement numbers are the author's measurements, not vendor benchmarks, and API usage still needs a TypeSafe key.
Questions about this entry
Short answers drawn from the linked project or page.
The three System One primitives: Noul for calibrated yes/no judgements, Choice for one of several options, and Score for a value on a number line. You send a state, any JSON value, plus the questions, and get back typed answers with probabilities.
Save it to ~/.typesafe.apikey or set the TYPESAFE_API_KEY environment variable. The library reads a default file or environment through ApiKey.fromDefaultFile() and ApiKey.fromEnv(), and both the library and CLI pick it up automatically.
Yes. The cli module builds a self-contained uber-jar, and --min sets per-question thresholds with exit code 1 when a threshold is not met, so a shell script or Jenkins job can wire it in as a pass/fail check; stdout stays silent unless you ask it to print.
The README documents local models such as CLEF and LAYA running through ONNX Runtime behind the same TypeSafeClient, with measured memory and per-request times on an Apple M5. The API path still needs a key; the local path is measured separately.
It removes the map-and-cast boilerplate. You declare a Java record whose fields carry @Noul (or choice and score) annotations describing the questions, and evaluateTyped decodes the response directly into the record.
The README's table compares each local member against the real jev-1.13.0 on 104 cached requests, reporting agreement on the same side of 0.5 for yes/no questions and matching choices; for example, Laya fp32 is listed at 85% and 64%.
Java 21. The project publishes core, mapping, and cli modules on Maven Central under io.github.dfa1.typesafe-java, and the cli uber-jar under the all classifier is the runnable artifact.
More Jev resources like this
A native Swift implementation of OpenJev, the Jev-compatible typed-decision server: send a state and typed questions over a Jev-compatible HTTP API and read every answer from model probabilities, on Apple silicon Macs or inside iPhone and Mac apps.
Jev Repos & SDKs#community#catalog-import#cobanov
Community
algorythm-canada/openjevswiftSwift
algorythm-canada/openjevswiftSwiftProbabilistic control flow for TypeScript: Hunch asks a System One model such as Jev yes/no questions, thresholded booleans, picks and ratings, and returns typed answers you can branch on in ordinary code.
Jev Repos & SDKs#community#catalog-import#cobanov
Community
carldaws/hunch-tsTypeScript
carldaws/hunch-tsTypeScriptA Java and Python implementation of parallel constrained decoding: fill a whole JSON schema of booleans and enums in two forward passes, with calibrated confidence per field and output that is schema-valid by construction.
Jev Repos & SDKs#community#catalog-import#cobanov
The Go server, agent CLI and MCP for Laya - open horizontal System 1 decision models that score structured features and return ranked pattern probabilities instead of text. It ships a JEV compatibility API and a migration guide.
Jev Repos & SDKs#community#catalog-import#madewithjev
Community
neko233-com/laya-goGo
neko233-com/laya-goGoTyped semantic decisions for TypeScript: ask Zod 4 or native question batches about your data and get validated answers through async/await or Web Streams, with confidence gates, retries, replay and observability. Jev runs via TypeSafe AI or Vercel evaluation.
Jev Repos & SDKs#community#catalog-import#logicrw
Community
nitoba/questionsTypeScript
nitoba/questionsTypeScriptPython with a smart if: .jev files add the ~ question operator and jev/case blocks that Jev answers with a number or a label, never text. Files are rewritten to plain Python at import time; every ~ is one paid TypeSafe API call.
Jev Repos & SDKs#community#catalog-import#logicrw
Community
sumanmichael/jevlangPython
sumanmichael/jevlangPythonFrom 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.