A dependency-free Python tool that rebuilds a site's internal link map with Jev. Code crawls and searches; Jev answers the link and anchor questions; a separate editor pass lets a deep model keep or cut the links Jev placed.

How does Jev judge whether a link belongs?

One request per page carries the copy once and asks 30 questions: a yes/no probability for each of 15 candidate targets and a multiple-choice anchor among phrases already in the copy. About half a second and $0.0005 per page in the recorded run.

What did the recorded site run cost?

The README's run: 566 pages and 8,460 link decisions in 5.95 seconds for $0.27, versus about $67 for Claude Opus 5 on the same queue with the same rubric. Jev placed 679 links on 334 pages and left 232 pages as they were.

How do I run System 1 on my own sitemap?

Python 3.11+ with no dependencies: export JEV_API_KEY (an OpenRouter key by default, or a TypeSafe key with JEV_URL=https://api.typesafe.ai/v1/systemone) and run python3 -m jev_linkmap.site https://your-site.com/sitemap.xml --mode system1.

What is the System 2 referee loop?

A deep model referees small page blocks with no clock, decides who was right where it and Jev disagreed, and rewrites the rubric's question wording, lessons, and thresholds; each version is scored on held-out pages, and Jev then maps the whole site with the best rubric.

What does the verify pass do before links ship?

python3 -m jev_linkmap.verify reads only the links Jev placed, twenty per call, and keeps or cuts each one without ever adding a link or changing an anchor. In the site run it read 679 links for $2.07 and kept 287.

What are the limits of the race comparison?

The comparison covers one site and one date, Jev's per-page cost is the README's own accounting, and the System 2 referee is described as a strong reader rather than ground truth. The manifest also shows NOASSERTION for the license.

Jev directory

CommunityJev Tools & CLIs

Jev Linkmap

A dependency-free Python pipeline that rebuilds a site's internal link map with Jev: code picks candidate targets and in-copy anchor phrases, Jev judges them, and only links that clear the threshold are placed.

See highlights Open github.com

Stars: 22
Language: HTML
Reading time: 3 min
Published by: Community
Author: stas4000
Last verified: Oct 2026

stas4000/jev-linkmap

User repository on GitHub

View on GitHub

Rebuild a site's internal link map in seconds with Jev, race Claude Opus 5 on the same queue, and let a deep model rewrite the rubric from the disagreements.

Forks: 0
Last pushed: Updated Oct 2026

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

Tagscommunity catalog-import logicrw seo internal-linking

Works forJev for Classification Jev for Decision gating

Used inJev for Software & DevTools Jev for Sales & Marketing

Languages: Python
Platforms: CLI OpenRouter
Capabilities: Open sourceRequires API key

Highlights

One request per page carries the copy once and asks 30 questions: a yes/no link probability and an anchor choice for each of 15 candidate targets.
In the recorded run: 566 pages and 8,460 decisions in 5.95 seconds at $0.27 total, versus about $67 for Claude Opus 5 on the same queue.
Anchors are phrases already in the author's copy, text that is already a link is never offered, and at most 3 new links land per page.
System 2 has a deep model referee disagreements and rewrite the rubric; two passes moved anchor agreement from 71% to 88% for $15.51 once.
A verify pass reads only placed links, 20 per call, and keeps or cuts each; on the site run it read 679 links and kept 287.

Quickstart

bash

export JEV_API_KEY=sk-or-...
python3 -m jev_linkmap.site https://your-site.com/sitemap.xml --mode system1

Watch out

Copy reviewed against the repo README and its run files on 2026-10-15. The cost race covers one site on one date, and the System 2 referee is described as a strong reader rather than ground truth. The GitHub manifest shows NOASSERTION for the license.

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.

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 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

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

Jev Linkmap

Highlights

Quickstart

Questions about this entry

What is jev-linkmap?

How does Jev judge whether a link belongs?

What did the recorded site run cost?

How do I run System 1 on my own sitemap?

What is the System 2 referee loop?

What does the verify pass do before links ship?

What are the limits of the race comparison?

More Jev resources like this

From the Jev guides

From the Jev AI model community