User repository on GitHub
A spatial reference explorer for creators. Local Jev query choices, metadata highlights and source-linked collections.
- Forks
- 4
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Cookbooks
RefGarden
A local 3D reference gallery where Jev chooses search phrases and highlights matching Met, NASA, Cosmos, and Prelinger Archive items from their text metadata.
- Stars
- 35
- Language
- TypeScript
- License
- MIT
- Reading time
- 3 min
- Published by
- Community
- Author
- AlbionaHoti
- Last verified
- Sep 2026
About this project
A space to find your next visual. Find images and short video references from one prompt. Explore artwork from The Met, space and science images from NASA, design references from Cosmos, and short films from Internet Archive's Prelinger Archives in a dark, three-dimensional gallery. Run locally · Try a prompt · Short videos · How Jev works · FAQ · Contribute.
Quoted from the project's README. Read the full README.
Works forJev for Ranking
Used inJev for Software & DevTools
Highlights
- Jev picks the search phrases and highlights references from titles and descriptions; it receives no image pixels, video frames, or audio.
- Image collection targets roughly a third each from The Met, NASA, and Cosmos, with archive clips counted separately.
- Clips must be MP4 files under three minutes and 80 MB; up to eight previews play at once, taking turns every 12 seconds.
- Duplicate filtering groups known size and format variants and matching Met records, but does not compare pixels.
- The hosted preview uses keyword retrieval without Jev; only the local Explore path calls TypeSafe.
Watch out
MIT-licensed, and the license covers the code rather than the media it finds. Needs Node.js 22 or newer and a TypeSafe API key, and the project installs its own pinned Bun runtime.
Questions about this entry
Short answers drawn from the linked project or page.
Only the metadata. Jev chooses the search phrases and reads image titles and descriptions to pick at most one highlight per image source. It receives no image pixels, video frames, or audio, does not watch clips or transcribe speech, and code handles retrieval, file eligibility, and playback.
Not the local Explore path. It expands Connections, takes your key, makes a small verification request, and saves it in the local ignored .env file, or you can set TYPESAFE_AI_API_KEY from .env.example before starting. The hosted preview accepts no Jev key and falls back to keyword retrieval instead.
Four documented reasons: source availability, the prompt itself, duplicate filtering across discovery rounds, and the roughly even split between The Met, NASA, and Cosmos, with Cosmos capped at a third of fresh image results. Clips need an eligible MP4 under three minutes, and missing matches leave uneven shares that the app reports.
No. The license covers RefGarden's code, not the media, and the README tells you to check each original item's rights and credits. Images arrive through The Met's Open Access collection API, where the adapter requests public-domain items, and video footage streams from the Prelinger Archives rather than being bundled.
Install Node.js 22 or newer, then npm ci, npm run build and npm start, and open http://127.0.0.1:4318. The project installs its own pinned Bun runtime, and a global Bun install is optional.
The first image batch requests up to 100 and later batches up to 30; discovery continues until Stop, three empty rounds or an error. Images target roughly a third each from The Met, NASA and Cosmos.
Short clips from the Prelinger Archives on Internet Archive, limited to MP4 files up to three minutes and 80 MB. Each batch examines up to 36 items and adds up to 12 eligible clips, with up to eight previews playing at once.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: A visual reference finder
X: LinkedIn bookmarks search
X: 1,066 reaction videos, searchable
X: Intent-based search in a Chrome extension
More Jev resources like this
A multiplayer demo workflow builder for Jev built with Liveblocks: wire Jev and LLM steps together, call the workflow through a REST API, and preview test runs with shared state and cursors.
Jev Cookbooks#community#typescript#demo
A demo application that navigates a Neo4j graph one hop at a time: Jev gets the outgoing relationships as a Choice plus a goal Noul in one system_one call, and beam search follows the probabilities.
Jev Cookbooks#community#python#neo4j
Community
jexp/neo4jevJupyter Notebook
jexp/neo4jevJupyter NotebookA small demo app that fetches a URL, turns the page into clean Markdown, and asks Jev ten bounded Choice questions about what a first-time visitor is likely to understand.
Jev Cookbooks#community#typescript#demo
A writing checker that runs seven named Jev questions, covering hedging, filler phrases, passive voice, tone, and actionability, and shows each verdict with its own confidence.
Jev Cookbooks#community#typescript#writing
Community
TypeSafeAI/clarity-judgeTypeScript
TypeSafeAI/clarity-judgeTypeScriptA terminal Wikipedia link-race tool: Chromium fetches article pages, Jev ranks their outgoing links by expected remaining hops, and Python runs a beam search without a search API or a controller model.
Jev Cookbooks#community#python#cli
Community
komikat/jev-bfsPython
komikat/jev-bfsPythonCloudflare Worker and React app that sends pasted corporate prose through about 45 typed Jev questions, then emits rustc-style diagnostics for passive-aggression, urgency and information density.
Jev Cookbooks#community#typescript#cloudflare
Community
asfarsadewa/human-compilerTypeScript
asfarsadewa/human-compilerTypeScriptFrom 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.