What is Sieve and what problem does it solve?

Point Sieve at a folder of Markdown — research notes, briefs, a blog, a knowledge base. crosscheck compares every note with every other and lists pairs that contradict each other or repeat the same point; ask puts one question to every note and ranks the folder by the answer.

How do crosscheck and ask differ?

crosscheck judges every pair of notes from both sides, reporting contradictions, repeats and notes at odds with several others. ask puts the same plain-English question to every note and ranks them by probability; a 1,140-item notebook came back in 2–3 seconds for about 2.5 cents.

Do I need embeddings or an index?

No. The README says there is no index to build and nothing to embed: Sieve is just Node 20.6+ with no dependencies, no build step and no database, and your notes stay as files.

What does a run cost?

Every run is estimated before it spends anything, and crosscheck stops at a budget. The README's example judged 60 items in 240 requests for about $0.05, and the Aesop fables example asked all 313 fables one question in 20 requests for $0.0035.

Which model does it use?

Jev — typesafe/jev-1.13 — through OpenRouter, or a compatible model on your own machine. Every answer comes back as a typed answer with a probability and Jev never writes text; Sieve ranks, thresholds and hands you the uncertain middle.

Cost and timing numbers are the README's own example runs rather than third-party benchmarks, and the pasted README truncates inside the typed-question table. A contradiction between an old note and a new one often means the old note is stale rather than the new one wrong.

Jev directory

CommunityJev Tools & CLIs

Sieve

A zero-dependency Node tool that finds where a folder of Markdown notes contradicts itself: crosscheck judges every note pair from both sides, and ask puts one plain-English question to every note and ranks the folder by probability.

See highlights Open github.com

Stars: 1
Language: JavaScript
License: MIT
Reading time: 3 min
Published by: Community
Author: nothans
Last verified: Oct 2026

nothans/sieve

User repository on GitHub

View on GitHub

Find where a folder of Markdown notes contradicts itself, and ask it anything. Every pair of notes crosschecked, every note asked, with Jev via OpenRouter.

Forks: 0
Last pushed: Updated Oct 2026

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

Tagscommunity catalog-import logicrw openrouter markdown

Works forJev for Verification Jev for Ranking Jev for Decision gating

Used inJev for Software & DevTools

Languages: JavaScript
Platforms: CLI OpenRouter
Capabilities: Open sourceRuns locallyRequires API key

Highlights

Just Node 20.6+, no dependencies, no build step, no database — notes stay files, with no index to build and nothing to embed.
crosscheck judged 60 items as 1,770 pairs from both sides: 3,540 judgments in 240 requests for about $0.05 in the README's run.
ask put one question to 1,140 notes and ranked them in 2–3 seconds for about 2.5 cents.
Also related, compare and narrow views, lint rules with a CI exit code, and a route/eval mode that measures accuracy against labels you have.
Runs against Jev through OpenRouter or a compatible model on your own machine; every run is estimated before it spends and crosscheck stops at a budget.

Quickstart

bash

node sieve.cjs crosscheck --config examples/research-notes/crosscheck.config.json

Watch out

Copy reviewed against the truncated repo README on 2026-10-15; the excerpt stops inside the typed-question table. Cost and timing figures come from the README's own example runs, not independent benchmarks, and the linked homepage blog post was not reviewed for this entry.

Questions about this entry

Short answers drawn from the linked project or page.

From the Jev guides

Original write-ups on what this entry is for.

Jev at home: voice pipelines, Assist routing, and sentinels

Oct 16, 2026

How Home Assistant conversation agents, macOS voice pipelines, and an advisory sentinel put a fast typed decision in the loop — one call per utterance, thresholds in code, and a daily token budget.

Jev Tools & CLIs#home-assistant#voice#automation

Read article

Search visibility, audited by decisions

Oct 16, 2026

Jev-based SEO tools split the crawl from the call: deterministic crawlers count and parse, while typed judgments rank fixes, place internal links, score search results, and track AI-answer mentions — with costs and caveats attached.

Jev Tools & CLIs#seo#search-visibility#audits

Read article

Jev on the Gateways: Vercel, Workers AI, and OpenRouter

Oct 12, 2026

The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.

Jev Tools & CLIs#gateway#vercel#cloudflare

Read article

From the Jev AI model community

Posts from builders shipping with Jev right now.

Follow @typesafeai

Sieve

Highlights

Quickstart

Questions about this entry

What is Sieve and what problem does it solve?

How do crosscheck and ask differ?

Do I need embeddings or an index?

What does a run cost?

How do I run Sieve on my notes?

Which model does it use?

What are its limits?

More Jev resources like this

From the Jev guides

From the Jev AI model community