What does the ego-jev skill do?

It completes browser tasks with Ego Lite driving a real Chromium browser while Jev actively resolves semantic choices: target selection, filtering, ranking, classification and text evidence judgments. An agent keeps the goal and the plan.

How do I install ego-jev?

Use the skills CLI: npx skills add ZHUBoer/ego-jev, or clone the repo into a skills directory like ~/.agents/skills/ego-jev. It is listed on skills.sh under zhuboer/ego-jev.

What goes to Jev versus staying in code?

Finite semantic selection, classification, per-item relevance and evidence checks default to Jev. Prices, arithmetic, counts, sorting, date comparison and exact matching stay in local code, and known actions need no inference.

What are the requirements?

Node.js 20 or newer (tested on Node 24), a working Ego Lite or ego-browser installation, and a TypeSafe API key. No npm packages are required because the helper calls the HTTP API directly.

Where does the API key live?

The helper reads TYPESAFE_API_KEY, then TYPESAFE_API_KEY_FILE, then ~/.config/ego-jev/api-key. The README recommends keeping the key outside the skill directory and running Jev from Node, never inside page.evaluate() or page.fetch().

What is the audit journal for?

An append-only local JSONL journal records calls with fields and limits, and can list, show and export reports as HTML, so you can review what the helper asked Jev and when, without sending it anywhere.

How does batching make it faster?

The README says speed comes from fewer planner turns and batched judgments rather than routing every browser action through an API call; exact matching in local code handles everything that does not need semantics.

Jev directory

CommunityJev Agents & MCP

Ego Jev — zhuboer

Agent skill that pairs Ego Lite's Chromium control with Jev's semantic judgment: the agent keeps the plan, Ego Lite drives the browser, and Jev resolves choices like which observed link actually matches the intent. Exact work stays in local code.

See highlights Open github.com

Stars: 2
Language: JavaScript
License: MIT
Reading time: 2 min
Published by: Community
Author: zhuboer
Last verified: Oct 2026

zhuboer/ego-jev

User repository on GitHub

View on GitHub

Complete browser tasks with Ego Lite and actively call Jev for semantic target selection, filtering, ranking, classification and text evidence judgments.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

This is one of several Jev listings for the same project. The canonical entry is Ego Jev — ZephyrDeng.

Tagscommunity catalog-import logicrw browser ego-lite skill

Works forJev for Classification Jev for Ranking Jev for Tool calling

Used inJev for Software & DevTools

Languages: JavaScript
Platforms: Browser
Capabilities: Open sourceRequires API key

Highlights

Listed on skills.sh and installable with npx skills add ZHUBoer/ego-jev; the agent keeps the goal while Ego Lite operates a real Chromium browser.
Jev resolves semantic choices: which observed link, button or card matches intent, categorization, ranking, and whether page text is evidence.
Prices, arithmetic, counts, sorting, date comparison and exact matching stay in local code, never in an inference call.
Ships a dependency-free Jev client (choose/evaluate/evaluateMany), an observe-decide-execute-verify workflow runner and a JSONL audit journal.
Requires Node.js 20+, an Ego Lite or ego-browser install and a TypeSafe key read from TYPESAFE_API_KEY, TYPESAFE_API_KEY_FILE or ~/.config/ego-jev/api-key.

Quickstart

bash

npx skills add ZHUBoer/ego-jev
git clone [email protected]:ZHUBoer/ego-jev.git ~/.agents/skills/ego-jev

Watch out

Copy reviewed against the truncated repo README on 2026-10-15; the excerpt ends inside the minimal semantic selection example. Two unrelated repos also named ego-jev (ZephyrDeng and jiangkoumo) already exist in this catalog, so check the repo field when comparing.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

X: Text-to-screenshot (Malay Vasa)

X: Google's MediaPipe Decision Maker, tested in the browser

X: Playwright CLI + Jev vs Playwright MCP

X: jev-ultrafast splits every browser decision into operation, target and value. JEV passed 18,000 stars in its…

From the Jev guides

Original write-ups on what this entry is for.

MCP tool routing: Jev in front of the tool call

Oct 16, 2026

Tool menus can outgrow the context that reads them. The catalog's Jev-based routers and gates narrow the list: benchmarked tool search, per-tool scoring, fail-closed approvals, and WebMCP in the browser.

Jev Agents & MCP#mcp#tool-routing#agents

Read article

Approval policies for coding agents: hooks, guards, and gateways