Can a probability really gate a build without constant flaking?

Only if you tune it. The digest recommends narrow literal questions, small single-domain targets, and thresholds calibrated on your own code, with probes that must pass intact and fail on deliberately broken copies.

Why not just ask a general model to review the diff?

General models return prose you must parse, with format drift and output billing. Here Jev returns a probability per question in the same request, so gating is numeric comparison with standard exit codes.

Will this choke on my large repository or partial commits?

Oversized targets stop with a setup exit code instead of being silently cut, and staged or diff modes check touched targets in full. Keep targets to one domain and retain a full CI run.

My specs use tables or another language; will checks still work?

English works best and other languages need threshold tuning on your own documents. Markdown tables are not sent to the model yet, so restate rows as lists or directly in the question.

What are jev-spec's size limits, and what happens if a target is too big?

A target is sent in one request, so more than 500 files, a file over 2 MiB, or more than 120,000 characters of code stops the run with exit code 2 instead of being truncated. Keep targets to one domain rather than the whole source tree.

How much does a jev-spec check cost and how fast is it?

Jev is priced on input tokens only, at $0.042 per million, and output is free, and every report prints an estimate. Its own full check of 9 targets and 29 rubrics runs in under four seconds for an estimated $0.001.

Can I validate my jev-spec setup without a TypeSafe key?

Yes. jev-spec check --dry-run validates the configuration offline without a key. Real checks need a TypeSafe API key, and in a diff or staged run a target none of whose files changed is skipped entirely.

CommunityJev Agents & MCP

Jev Spec

Pre-commit and CI gate that turns each Markdown requirement into one Jev yes-or-no question over the paired code and fails the build when probabilities miss configured thresholds.

See highlights

Open github.com

Stars: 11
Language: TypeScript
License: MIT
Published by: Community
Author: nozomi-koborinai
Last verified: Oct 2026

nozomi-koborinai/jev-spec

User repository on GitHub

View on GitHub

⚡ Catch spec drift on every commit: check your code against your Markdown specs with TypeSafe AI's Jev model.

Forks: 3
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.Homepage

About this project

🌐 日本語 | 简体中文 | 한국어. Catch spec drift on every commit. jev-spec checks your code against the requirements in your Markdown specification and fails the build when they drift apart. It asks TypeSafe AI's Jev model one focused question per requirement, gets a probability back, and compares it with a threshold you set. That makes it small enough for a pre-commit hook and strict enough for a CI gate. Example report. The layout is what the CLI prints (timing and cost lines omitted); the probabilities are illustrative.

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

Tagscommunity jevbest catalog-import awesomejev

Works forJev for Classification

Used inJev for Software & DevTools

Languages: TypeScript

Highlights

One request per target sends the spec, code, and every rubric together and asserts with numeric thresholds and exit codes.
Supports noul, choice, and score rubrics in a typed config with staged and diff modes for fast local loops.
Enforces workspace root jails, symlink checks, revision sanitizing, and size caps instead of truncating oversized targets.
Self-tested with per-requirement breaking probes that must pass intact and fail on the broken copy.

Watch out

A probability is not a proof: the digest advises tuning thresholds on your own code and keeping targets small and literal. 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

JevLoop

Runnable agent harness that moves pick, score and yes-no loop choices to Jev or Laya in milliseconds while reserving the language model for writing, compiling DECISION.md into executable questions.

Jev Agents & MCP#community#jevbest#catalog-import

Community

zjunlp/JevLoopTypeScript

jev-test-filter

Diff-aware test selector that sends the git diff to Jev with one question per test, then prints the native filter flags and hands them to vitest, jest, node, bun, Playwright, cargo, or go test.

Jev Agents & MCP#community#jevbest#catalog-import

Community

mizchi/jev-test-filterTypeScript

pi-jev-router

Pi extension that classifies each task with Jev over OpenRouter and routes it to a Pareto-optimal model balancing measured quality, cost, and latency under a role policy.

Jev Agents & MCP#community#jevbest#catalog-import

Community

philippdubach/pi-jev-routerTypeScript

patdown

Node CLI that lints a file tree against fuzzy natural-language rules in one Markdown file, judging each matched file with a swappable TypeSafe-backed judge.

Jev Agents & MCP#community#jevbest#catalog-import

Community

tyler-dot-earth/patdownTypeScript

jevcore

106

Decision layer exposing Jev's typed yes-no, choice, and score answers to DeepSeek Harness and MCP hosts, offline by default with every network transmission disclosed per feature.

Jev Agents & MCP#community#jevbest#catalog-import

Community

PerryLink/jevcoreTypeScript

pi-jev

Pi extension that pairs a local coding model with hosted Jev judgments for skill hints, file retrieval, and evidence ranking, with caching and failure bounds on every call.

Jev Agents & MCP#community#jevbest#catalog-import

Community

madeye/pi-jevTypeScript

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai