Does fallback mean Jev still made the move?

No. Fallbacks are visibly labeled and counted separately; the dashboard never presents a fallback choice as a successful Jev decision.

What does Jev actually return per turn?

A selected legal action with confidence and distribution over legal candidates plus latency, token, cost, and redacted metadata; no chat loop or invented reasoning display.

How do I avoid leaking credentials while streaming?

Keep .env local and ignored by Git, use a dedicated Showdown account, and rely on dashboard redaction while rotating exposed credentials immediately.

Why not judge it by win rate alone?

Historical artifacts are not presented as product or strength evidence; harness correctness and observability must be evaluated separately from playing strength.

What Jev endpoint and model does the agent use by default?

It sends typed requests to JEV_ENDPOINT, which defaults to https://opencode.ai/zen/v1/systemone, using JEV_MODEL jev-1.13-free and a JEV_AUTH_TOKEN that defaults to Bearer public.

What does the validator check before submitting an order?

Action identity, membership in the legal candidate set, response freshness, probability validity, and the expected request fingerprint. Only a valid, legal choice is submitted.

How do I run the live battle dashboard?

Run python -m jev_showdown.main serve --port 8000, then open http://localhost:8000 and select START JEV BATTLE. One live battle session runs at a time.

CommunityJev Demos & Experiments

Pokemon Showdown Jev Agent

Gen 9 Random Battles agent that sends a structured battle view to Jev each turn, validates the legal choice, and shows telemetry on a dashboard.

See highlights

Open github.com

Stars: 0
Language: JavaScript
Published by: Community
Author: NatBrian
Last verified: Oct 2026

NatBrian/pokemon-showdown-jev-agent

User repository on GitHub

View on GitHub

Autonomous Pokemon Showdown battle agent powered by Jev AI (System One Model by TypeSafe AI), with a validated poke-env harness, safe fallback decisions, and a live dashboard for transparent battle state, legal moves, probabilities, telemetry, and Jev decision history.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

The supported workflow is the live dashboard. A dedicated Showdown account connects to the public ladder, the agent observes a real battle, sends a structured decision request to Jev, validates the returned choice, and submits a legal order. The dashboard shows the complete observable path. For every battle turn, the agent: Receives the current battle request from poke-env. Enumerates every legal move and switch action. Computes deterministic, observable facts for each candidate.

Quoted from the project's README. Read the full README.

Tagscommunity github

Works forJev for Tool calling

Used inJev for Software & DevTools

Languages: JavaScript

Highlights

Enumerates legal moves and switches, computes observable facts, and preserves opponent fog of war in the Jev request.
Validates action identity, legality, freshness, and probabilities; illegal or timed-out turns use labeled fallback ladder.
Dashboard splits Live Showdown, Jev decision, and harness facts with inspect drawer for full technical record.
Needs Python plus uv, dedicated Showdown account, Jev endpoint token, and local .env kept out of Git.

Watch out

Live ladder only, not a strength claim; rate limits, network, variance, and hidden info affect runs. 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.

More Jev resources like this

jev-playground

Zero-dependency local playground showing six Jev workflows from triage to eval judging with shippable snippets.

Jev Demos & Experiments#community#github

Community

AviroopPaul/jev-playgroundJavaScript

ego-jev

Agent skill that ports the ultrafast browser loop into ego lite: ego executes while Jev picks the operation and target in one call per step, with escalation instead of guessing on risky actions.

Jev Demos & Experiments#community#awesomejev#catalog-import

Community

ZephyrDeng/ego-jevTypeScript

tev1-50-use-cases

Fifty runnable tev1 decision-model examples through local Ollama, from ticket triage and guardrails to six playable games, with a live request viewer and scoreboard.

Jev Demos & Experiments#community#github

Community

PromptEngineer48/tev1-50-use-casesPython

jev-browse

Browser agent where Jev makes every navigation decision and agent-browser performs clicks and typing, stopping when the named goal page is reached or blocked.

Jev Demos & Experiments#community#jevbest#catalog-import

Community

kyrylosyzonenko/jev-browseJavaScript

jev-browser-control

An MCP server that lets Claude or Codex drive a real Chrome window: Claude plans the task and Jev picks each click, keystroke and scroll from a numbered element list, with irreversible actions held for confirmation.

Jev Demos & Experiments#community#awesomejev#catalog-import

Community

nexibeo/jev-browser-controlJavaScript

jev-browser

Playwright wrapper that routes natural-language operations and assertions through Jev while clicks, typing and waits stay deterministic, with verification provenance.

Jev Demos & Experiments#community#jevbest#catalog-import

Community

tontoko/jev-browserJavaScript

From the Jev guides

Original write-ups on what this entry is for.

Per-Tick Decisions in Games

Sep 28, 2026

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

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai