Jev resources, sorted by the team that would run them.

The official docs, the SDKs, and the open-source repos worth your time, grouped for marketing, sales, support, and engineering. A shortlist, not a dump. Every link was checked on the date below.

By Alpesh Nakrani · Links checked September 27, 2026

5official docs

4official SDKs + skills

17community picks

3field guides here

Start with a guide

Jev quickstart: Make your first Jev call through Vercel AI Gateway: one key, one file, typed probabilities back.
SEO internal linking with Jev: Score every page against your money pages in one call each, then keep only the links an editor would.
Buyer-intent prospecting with Jev: Find people asking for what you sell, drop the vendors, and check every message fits before it goes.

01/Official docs and pricing

Official docs and pricing

Start here. Every number in the Jev guides on this site traces back to one of these pages.

Jev on Vercel AI GatewayModel ID, pricing, 32K context window, and a code sample.Vercel AI SDK: experimental_evaluateParameters, return shape, and errors for the evaluate call.AI SDK reference Vercel AI Gateway docsKeys, spend limits, and billing for the gateway route.Vercel TypeSafe documentationChoice, score, and true-or-false questions; confidence; patterns.TypeSafe Introducing System One models and JevThe launch post: what Jev is for, latency claims, launch pricing.TypeSafe · Sep 15, 2026

02/Official SDKs and agent skills

Official SDKs and agent skills

For calling TypeSafe directly, or giving your coding agent the conventions up front.

typesafe-ai/skillsAgent skills for building with the System One API.MIT · 2,258 stars typesafe-ai/typesafe-sdk-jsOfficial TypeScript/JavaScript library.MIT · 247 stars typesafe-ai/typesafe-sdk-pythonOfficial Python library.MIT · 237 stars typesafe-ai/system-one-adapter-pythonDrop-in client backed by LLM APIs, useful for comparing Jev with a general model.MIT · 310 stars

03/Marketing, SEO, and content

Marketing, SEO, and content

Page grading, link audits, and editorial gates. Pair with the internal-linking guide.

prantikmedhi/anchorlintEvidence-first internal-link auditing with optional Jev judgments and CI gates.MIT kitze/pagegradeBrowser extension that grades page sections for clarity, writing, and on-page SEO.MIT tjkimcloud/jev-site-auditorScores every page against your own brand and answer-engine rules.MIT RefoundAI/jev-editor-skillClaude Code skill that scores drafts for AI tells, voice, quality, and SEO.Agent skill stas4000/jev-marketingSeven marketing workflows with a Python CLI and MCP tools.MIT leepokai/jev-adrankAd ranking and creative review without a trained CTR model.MIT

04/Sales and prospecting

Sales and prospecting

Lead scoring and call review. Pair with the buyer-intent prospecting guide.

promptgtm-shared/clay-jev-people-rankerClay lead scoring and people-search ranking as an agent skill.MIT LiamSherline/jev-lead-scorerFit score, outreach angle, and skip flag, with a funnel that tracks which openers book meetings.MIT wsmoak/jev-sales-callsReads sales-call transcripts against a product capability library.Example

05/Support and operations

Support and operations

Triage and moderation, where a probability decides the queue.

soderlind/jev-comment-triageWordPress comment moderation for spam, scams, and toxicity, in the background.WordPress maraichr/jev-triageCross-border B2B case triage prototype.MIT

06/Integrations and tooling

Integrations and tooling

Put Jev inside the tools you already run.

jkudish/jev-mcpJev judgments as MCP tools for Claude and other agents.MIT · 412 stars superagents-lab/jev-searchWeb search with Jev for source selection and relevance ranking.MIT · 472 stars vibe-with-me-tools/n8n-nodes-jevn8n node to classify, route, and score, sending unsure items to review.MIT

07/Bigger lists

Bigger lists

When you want everything, not a shortlist.

yibie/awesome-jevCurated list of public projects, integrations, and discussions.1,822 stars walidboulanouar/awesome-jev-use-casesUse cases, demos ranked by likes, repos, limits, and cost notes.CC0 · 259 stars Broch Builds: Jev resources130+ demos, repos, and docs sorted by team, with view counts.Ben Broch

Judgment

How I would pick from this list

Most of these repos are days old. That is normal for a model that launched on September 15, 2026, and it means you should read the code before you trust the output. Three filters I would apply:

Does it show its thresholds? A tool that hides the probability behind a verdict hides the part you need to calibrate.
Does it keep a human step? Good Jev tools send the unsure middle band to review. That is the human-in-the-loop evaluation pattern, built in.
Can you run it on your own labelled examples? If not, you are trusting someone else's eval set.

FAQ

Jev resources: common questions

Where is the official Jev documentation?

TypeSafe publishes the model documentation at docs.typesafe.ai. If you call Jev through Vercel AI Gateway, the model page on vercel.com lists pricing and limits, and the AI SDK reference documents experimental_evaluate.

Is there an official Jev SDK?

Yes. TypeSafe maintains official JavaScript/TypeScript and Python SDKs on GitHub under typesafe-ai, plus a skills repository for coding agents. Through Vercel AI Gateway you can use the ai package instead.

Can I use Jev from Claude Code?

Yes. Install TypeSafe's agent skills so the agent knows the API, or add an MCP server such as jev-mcp to give Claude typed Jev judgments as tools.

How were these repos chosen?

Each one is public, does something a marketing, sales, support, or engineering team would actually run, and was checked on GitHub on September 27, 2026. Star counts are shown only where they help you judge adoption.