jevseo
Scores a page for SEO, answer engines and generative engines, using code for measurable checks and Jev rubric scores for judgment.
Overview
JevSEO scores a URL or draft for three different jobs: classic SEO (does the page serve the query intent and add something new), AEO (can a machine lift a correct, self-contained answer for snippets and voice) and GEO (is there anything worth quoting and enough provenance for AI assistants to cite). Measurable checks run with fixed thresholds in code, and only questions that need judgment go to Jev.
Each judgment is a Score question on a rubric whose levels are written out in the repository, for example whether a 40 to 60 word passage could stand alone as an answer. Jev returns a position on the rubric, a probability per level and a confidence, which the README argues is more defensible and repeatable than asking a chat model for a report. Output is three scores and a ranked list of fixes with evidence. The README estimates about two seconds and £0.0001 per page. It runs locally or via a hosted demo where your own TypeSafe key stays in the browser.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | Does a passage of 40 to 60 words answer the reader's question correctly if lifted out and shown on its own? | No passage survives being lifted out · Candidate too long or pronoun-dependent · Workable passage needing light trimming · Self-contained, correctly scoped passage |
| Score | Does this page serve the intent behind the query and say something other results do not?(inferred) | — |
| Score | Is there anything here worth quoting, with enough provenance to cite?(inferred) | — |
How it works
- Paste a URL or draft
- Code runs fixed-threshold measurements
- Jev scores judgment questions on written rubrics
- SEO, AEO and GEO scores are computed separately
- Ranked fixes with evidence are returned
Strengths
- Rubrics are in the repo and editable
- Separate scores for three different goals
- Low reported cost and latency per page
- Keys stay on your machine or in your browser
Caveats
- Rubric scores are model judgments, not ranking guarantees
- Requires a TypeSafe key
- Demo-stage project
Getting started
Try the hosted version at jevseo.epergaboni.com with your own TypeSafe key, or run it locally.At a glance
- Type
- Demo
- Status
- prototype
- Language
- TypeScript
- Runtime
- web app
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Typed SEO, AEO and GEO judgments powered by Jev, a System One decision model.
wh000wh000/awesome-jev-liveTyped SEO, AEO and GEO judgments powered by Jev, a System One decision model. Code owns the rules, the model owns the meaning.
hellogumbo/awesome-jev
Frequently asked questions
Why three scores?
Search ranking, answer-engine extraction and AI citation reward different things and sometimes conflict.
Why Jev instead of a chat model?
Jev returns a position on your written rubric with probabilities, which is more repeatable than a generated report.
What does a page cost?
The README estimates about £0.0001 and two seconds.
Related projects
- jev-seo★ 610
Live SEO audit from one homepage URL: crawl, 52 rules, Core Web Vitals and Jev page judgments, reported as PDF, XLSX and Markdown.
PluginbetaPython Site audit for SEO and AI-search visibility: a deterministic crawl, narrow Jev Choice, Score and Noul questions, and a live dashboard.
AppbetaTypeScriptChrome extension that grades readable page sections on ten rubric scores with Jev via Vercel AI Gateway and gives the page an A-E grade.
ProjectbetaTypeScriptGEO analytics platform tracking brand visibility in AI answers; a feature flag moves its visibility classifiers onto Jev booleans.
ProjectproductionTypeScriptAnalytics engine that rates published posts on pre-registered rubrics with Jev and tests ratings against real engagement.
ProjectbetaTypeScriptCodex plugin that runs polls on simulated reader profiles with a System One model, recording typed answers for follow-up studies.
ProjectbetaTypeScript