Skip to content
JevDirectory.orgTraffic count unavailable. View site traffic

Jev directory

CommunityJev Practices & Patterns

Poorjev

An open, local-first alternative to Jev that reproduces the Choice, Score, and Noul interface on commodity zero-shot NLI models with temperature scaling and conformal abstention, and runs offline with no API key.
See highlightsOpen github.com
Stars
10
Language
Python
License
MIT
Reading time
3 min
Published by
Community
Author
rupeshpoojary9
Last verified
Sep 2026

About this project

The poor man's Jev. An open source, local-first "System One" decision layer for LLM apps: typed decisions with provably calibrated confidence . No API key. No waitlist. Your model's 0.9 is a vibe. poorjev's 0.9 is a measurement. Every LLM-in-JSON-mode hands you a confidence score and hopes you don't check it. poorjev checks it. On the shipped eval set it cuts calibration error (ECE) from 0.170 to 0.071 with zero loss of accuracy, and it runs on your laptop with no API key.

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

Tagscommunitypythoncalibrationevaluationmcpclassification
Works forJev for EvaluationJev for Tool callingJev for Scoring
Used inJev for Software & DevTools

Highlights

  • Cuts calibration error (ECE) from 0.170 to 0.071 on its shipped eval set with zero loss of accuracy.
  • Temperature is fit by 5-fold cross-validation, so the after number is measured on held-out data.
  • Reproduces Jev's Choice, Score, and Noul interface on a local zero-shot NLI model in one batched pass.
  • Ships an MCP server exposing gate, judge, classify, rate, and decide tools to Claude Code with no API key.
  • Its crossbench harness measures Jev, von, Laya, and poorjev; it leads open options on the mixed-primitive set but not Banking77.

Quickstart

python
from poorjev import Client, Choice, Score, Noul
client = Client()
result = client.ask(
    state="I've emailed three times and I'm still being double-charged.",
    questions={
        "topic": Choice(["billing", "technical", "account"]),
        "frustration": Score(levels=["low", "medium", "high"]),
        "is_urgent": Noul("The customer needs a response today."),
    },
)
print(result["topic"].value, result["topic"].confidence)

Watch out

MIT-licensed; runs on CPU offline after one model download with no API key or GPU, and the README states it does not beat Jev and still trails on the high-cardinality benchmark.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

X: Jev in front of the HeyGen MCP

X: Playwright CLI + Jev vs Playwright MCP

X: A local MCP server that cuts computer use from 23s to 0.8s. Cartwright built Cortex, a local MCP server where…

X: A side panel that drives any site's WebMCP tools

From 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.

Follow @typesafeai