User repository on GitHub
A natural-language task workspace comparing performance across AI models (TypeSafe's Jev, GPT-5.6 Luna, and Gemini 3.8 Flash)
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Demos & Experiments
Tempo Jev Demo
A local natural-language workspace for a fictional team: ask for a board, table, cards, bubbles, grouped canvas or calendar, then follow up to reassign work, move panels, add events or build a checklist.
- Stars
- 2
- Language
- TypeScript
- License
- MIT
- Published by
- Community
- Author
- mychaelangelo
- Last verified
- Oct 2026
About this project
Tempo is a local, natural-language task workspace for a fictional team. It can use TypeSafe's Jev, OpenAI Luna, or Gemini Flash to turn your requests into workspace changes. Ask it to show work as a board, table, cards, bubbles, grouped canvas, or calendar; follow up to change the same view, complete or reassign tasks, schedule local events, capture notes, or edit a personal checklist.
Quoted from the project's README. Read the full README.
Works forJev for Classification
Used inJev for Software & DevTools
- Languages
- TypeScript
Highlights
- Runs TypeSafe Jev, OpenAI Luna and Gemini Flash against the same request, with per-session timings in the right panel.
- No database or seed step: 19 fictional tasks and 11 meetings live in one fixture file behind a clock fixed for reproducibility.
- Workspace changes and Undo history are in memory for the current browser page; Reset or a refresh returns the initial data.
- Browser journeys run against intercepted API responses after `npx playwright install chromium`, so they need no provider keys.
Watch out
The author built it with Codex and says he has not personally reviewed all of the code, calling it an experimental demo that may contain bugs. The speed and cost figures come from his own local test on matching requests using published uncached token prices, not from a controlled benchmark. Everything is in memory for the current page, so a refresh discards your workspace. There is no authentication, the README explicitly says it is not ready to expose publicly, and prompts plus workspace context go to whichever external provider you pick. The request and spend guards are estimates rather than billing ceilings, since a timed-out call can still bill. Copy reviewed against the repository README on 2026-10-04, not from running it.
Questions about this entry
Short answers drawn from the linked project or page.
It is the author's own local test on requests he considered matching, using published uncached token prices. The README itself says live results and latency vary, and the app shows you timings for the current page session so you can compare providers on the same request by submitting it again before changing the workspace. Treat the ratio as a starting hypothesis, not a benchmark result.
It is gone. Workspace changes and the Undo history live in memory for the current browser page only, and refreshing or pressing Reset returns you to the initial fixture data. There is no database and no seed step, so nothing is persisted. The fictional clock is fixed to keep relative-date requests reproducible, which is also why the demo cannot be pointed at real data as-is.
Not as shipped. There is no user authentication, the API can incur paid provider usage, and the README says it is not ready to be exposed publicly without authentication, access controls and operational safeguards. Prompts and the current workspace context are sent to whichever external provider you select, so real or confidential material leaves the machine. Provider keys belong in server-side variables, never VITE_ ones.
The port. Vite proxies /api to 127.0.0.1:8787, so that port has to be free, and if you change PORT in .env you also have to update the proxy in vite.config.ts. You need Node 22 or newer. With no live provider configured the only option is a local Preview provider, which is a limited deterministic fallback rather than a model, so it will not interpret natural language properly.
TypeSafe Jev, OpenAI Luna, or Gemini Flash. Each is configured from .env, and with no live provider the limited deterministic Preview provider is available only as an installation check.
The server defaults to 400 interactive requests and an estimated $5, a 15-second request timeout, and port 8787. The README notes these guards are not hard billing ceilings.
Submit the same request with another provider before changing the workspace. Tempo sends that provider the saved pre-request state, so both models start from the same data.
More Jev resources like this
98GitHub stars
Self-hosted Docker service and web UI that packages Laya structured-decision inference behind a compatible System One endpoint.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
1Panel-dev/laya-serverPython
1Panel-dev/laya-serverPython20GitHub stars
General-purpose System One agent where Jev chooses tools and arguments while Jot executes them and loops until it can reply.
Jev Demos & Experiments#community#awesomejev#catalog-import
Community
runta-dev/jotTypeScript
runta-dev/jotTypeScript23GitHub stars
Browser drone-fleet lab where TypeSafe Jev steers each drone as fast reflexes and an optional planner offers one-use strategy advice when confidence drops, without pausing control.
Jev Demos & Experiments#community#jevbest#catalog-import
23GitHub stars
Chrome extension that sends each X post to Jev with six typed questions in one call near the viewport, then renders a behavior chip under the post plus a session cost and latency panel.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
oso95/x-scannerTypeScript
oso95/x-scannerTypeScript14GitHub stars
Manifest V3 Chrome extension in TypeScript that sends each LinkedIn post to Jev and collapses bait, self-promo, off-topic, and sponsored posts into a one-line bar with reasons and scores.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
stefw/lkcleanTypeScript
stefw/lkcleanTypeScript11GitHub stars
Standalone Flue example that routes support chat through Cloudflare Workers AI, using gpt-4o-mini for chat and Jev for typed intent and urgency routing.
Jev Demos & Experiments#community#awesomejev#catalog-import
Community
matthewp/flue-jev-demoTypeScript
matthewp/flue-jev-demoTypeScriptFrom the Jev guides
Original write-ups on what this entry is for.
A game tick is a state plus a legal action list. How Clash Royale, Minecraft, and the earlier Mario, Pokémon, StarCraft, and drone loops ask Jev one step at a time.
Jev Demos & Experiments#gaming#agents#control
Read article
From the Jev community
Posts from builders shipping with Jev right now.