User repository on GitHub
Researching how Jev can accelerate tasks that rely on LLM decision-making.
- Forks
- 2
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Agents & MCP
Jev-in-the-Loop
A research repository exploring which LLM-decision tasks Jev can accelerate, starting with a Codex plugin: Codex understands the goal and prepares inputs, Jev picks the next browser action, and the plugin executes it in local Chrome.
- Stars
- 4
- Language
- Python
- License
- MIT
- Reading time
- 2 min
- Published by
- Community
- Author
- tongyun1
- Last verified
- Oct 2026
Highlights
- Browser is the first module: a Codex plugin where Codex prepares inputs and Jev chooses the next action in one TypeSafe request.
- Demo flows include searching Beijing hotels on Ctrip and stopping at the booking form, and finding Stanford CS336 on Bilibili.
- Your existing Codex setup plus a TypeSafe API key is all the model setup needed — no additional text-generation service.
- Searches, clicks, typing and navigation flow step by step in local Chrome, with action, target and prepared input chosen together.
- Python 3.12+, MIT license and self-described Alpha status; the Codex plugin lives in plugins/jev-browser-plugin.
Watch out
Copy reviewed against the truncated repo README on 2026-10-15; the capture ends mid-sentence in the workflow section, so install steps, benchmarks and pricing are not documented here. The only requirement captured is an existing Codex setup plus a TypeSafe API key.
Questions about this entry
Short answers drawn from the linked project or page.
A research repository investigating how Jev can accelerate tasks that rely on LLM decision-making, and which decisions Jev can take on to make a whole task faster. Browser, a Codex plugin, is its first module.
Codex turns your request into a plan and prepares the words to type; Jev then chooses the next action, its target and the prepared input in a single TypeSafe request; the plugin executes that in local Chrome.
Two documented flows: finding Beijing hotels for specific dates on Ctrip, choosing a room and opening the booking form for the human to confirm, and searching Bilibili in English for Stanford CS336 and opening the first video.
Python 3.12+ and an existing Codex setup plus a TypeSafe API key — the README says no additional text-generation service is needed. The project is MIT-licensed and self-described as Alpha status.
No. The design keeps the assistant that understands your task — Codex — and gives its browser work a fast decision loop through Jev. The README frames this as an upgrade to the workflow you already use.
The README's differentiator is the division of labor: Codex handles context, planning and prepared input, while Jev makes the per-step choice without a separate text-model API key. It is framed as research into which decisions Jev speeds up, module by module.
The repository belongs to tongyun1, created 2026-09-21 and last pushed 2026-09-22, with a contribution guide linked. The stated direction: browser first, with more modules growing from the decision loops explored next.
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…
More Jev resources like this
Codex Agent Skill that hands continuous browser work to Jev: one sentence starts a task, Jev picks the actions and targets, and Codex replies once the run completes. Reuses your own Chrome and reads the DOM instead of screenshots.
Jev Agents & MCP#community#catalog-import#logicrw
Community
wanghai673/jev-browser-skillPython
wanghai673/jev-browser-skillPythonAn MCP server that lets Claude hand a whole browser task to Jev: Claude plans and verifies while Jev picks each click, type or select from a numbered table of visible controls in about 100 ms, using browser-use/jev-ultrafast's loop.
Jev Agents & MCP#community#catalog-import#madewithjev
Community
msalvalaggio/jev-reflexPython
msalvalaggio/jev-reflexPythonLets AI agents use Windows applications through MCP or the command line: hand off a bounded task and Jev observes windows, chooses controls, types supplied text and navigates locally, returning a result or a request for help.
Jev Agents & MCP#community#catalog-import#logicrw
Community
jacks3tr/jev-desktopPython
jacks3tr/jev-desktopPythonAgent 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.
Jev Agents & MCP#community#catalog-import#logicrw
Community
zhuboer/ego-jevJavaScript
zhuboer/ego-jevJavaScriptAn open-source, self-hosted AI agent platform: role-based agents with custom tools, RAG knowledge bases, and 50+ integrations, deployable to Slack, WhatsApp, Teams, Telegram, a web widget, or a REST API from one platform.
Jev Agents & MCP#community#catalog-import#logicrw
Community
getsynkora/synkora-aiPython
getsynkora/synkora-aiPythonJob-application skill with a memory for CLI agents: given a résumé and a posting it fills what your record supports, drafts grounded prose when you allow it, and leaves the rest open rather than guessing.
Jev Agents & MCP#community#catalog-import#logicrw
Community
theadaply/jev-applyJavaScript
theadaply/jev-applyJavaScriptFrom 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.