User repository on GitHub
Jev (TypeSafe) ile doğal dille kıyafet arama: karar Jev'de, sıralama kodda
- Forks
- 1
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Demos & Experiments
Jev Kiyafet Bul
Turkish-language natural-language search over an outerwear catalog: Jev reads each request in one call and scores candidate matches, while code handles price and size regex, filtering, ranking and the results UI — with a side panel showing what Jev decided.
- Stars
- 4
- Language
- TypeScript
- Reading time
- 3 min
- Published by
- Community
- Author
- selmakcby
- Last verified
- Oct 2026
- Languages
- TypeScript
- Platforms
- Next.jsVercel AI Gateway
- Capabilities
- Open source
Highlights
- One Jev call reads the request — scope, type, length and color as Choice, plus belt, hood, waterproof and budget as Noul.
- A side panel replays each search step by step with real data, so you can see exactly what Jev decided, score by score.
- Scope gate: when out-of-catalog confidence reaches 0.5, scoring is skipped and the app returns an empty result with an explanation.
- Code narrows the catalog to at most 60 candidates, Jev scores matches in batches of 10, and ranking is score × confidence, top 12.
- Without a key the app runs on a deterministic fake Jev; tests run against the fake, and JEV_SAHTE=1 forces it even with a key.
Quickstart
bash
npm install
npm run dev # http://localhost:3000
npm test # vitest, run against fake JevWatch out
Copy reviewed against the truncated repo README on 2026-10-15; the capture ends inside the commands section, so the data-prep scripts and full setup were not reviewed. The ~300ms median latency is the author's own video measurement, and the README states the architecture is not disclosed — commentary on Jev's internals there is inference.
Questions about this entry
Short answers drawn from the linked project or page.
A Turkish-language demo that turns a sentence like 'a long beige trench, belted, size M, under 3000 TL' into filtered, ranked outerwear results, with a dark side panel showing what Jev decided at every step.
In two stages: one Jev call parses the request into Choice fields (scope, type, length, color) and Noul flags (belt, hood, waterproof, budget), then Jev scores each candidate's fit on an ordered scale. Code does the filtering and ranking.
Jev does not read numbers here: the budget Noul only answers whether a limit exists, and regex extracts the actual price ceiling and size. Filters below 0.5 confidence are not applied, and size is not filtered for knitwear.
If the out-of-catalog confidence is at least 0.5, the app skips scoring entirely and returns an empty result with a message saying the catalog only holds outerwear and knitwear. The panel marks those steps as skipped.
No. Without a key the project runs a deterministic fake Jev whose answers are shaped correctly but meaningless, and the panel shows a fake badge. For real decisions use a TypeSafe key (waitlist) or Vercel AI Gateway with model typesafe-ai/jev.
Run npm install then npm run dev and open http://localhost:3000. npm test runs the vitest suite against the fake Jev; the catalog build scripts pull data via Apify and are documented as a paid, one-time step.
The README is in Turkish, the demo video is marked coming soon, and the README itself says Jev's internal architecture is not disclosed, so any explanation of how it works inside is a guess. The repo was created 2026-09-28.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: LinkedIn bookmarks search
X: 1,066 reaction videos, searchable
X: Intent-based search in a Chrome extension
X: Margin, bookmark search
More Jev resources like this
Vercel Labs example that routes three forms by context with Jev and AI SDK: Jev's team choice is accepted when its confidence is at least 95%, otherwise openai/gpt-6-luna-fast independently evaluates the submission and its choice becomes final.
Jev Demos & Experiments#community#catalog-import#logicrw
A Next.js and MUI showcase of Jev classifying Enron-Spam emails: one call asks a Noul spam question, a Choice category question and a Score urgency question at once. Includes a dataset view, a no-key example tab and a stand-alone Python script.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
harsha89/jev-demoTypeScript
harsha89/jev-demoTypeScriptJevSpeak explores whether a model that cannot generate text can still talk: about 13 parallel Jev decisions per turn become a typed Semantic IR that a deterministic compiler renders in English or Chinese. Live at jevspeak.org.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
mm-sheng/jevspeakTypeScript
mm-sheng/jevspeakTypeScriptAn interactive Next.js workbench for Jev's three decision primitives through OpenRouter: rank 95+ categories with one Choice pass, test binary Noul decisions with transfer curves, and grade inputs on ordinal Score rubrics.
Jev Demos & Experiments#community#catalog-import#logicrw
A BYOK workbench for practicing Jev concepts without official access: emulates State, Noul, Score and Choice via Groq, OpenAI, Anthropic, Gemini or local Ollama, or runs against native Jev with authorized credentials.
Jev Demos & Experiments#community#catalog-import#logicrw
A solo mahjong coach that takes a 14-tile hand and has Jev recommend which tile to discard, with efficiency, safety and wait scores plus a Japanese explanation, beginner-to-advanced lesson tracks, and a practice mode against three CPUs. Progress and consultations live in Turso.
Jev Demos & Experiments#community#catalog-import#madewithjev
Community
smilior/kiru-hai-coachTypeScript
smilior/kiru-hai-coachTypeScriptFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.