Jev Browser Control

Let Claude or Codex drive a real Chrome. You ask in plain words; Claude plans, and Jev, TypeSafe's decision model, picks each click, keystroke and scroll in about half a second for a fraction of a cent.

Created by Jeroen Erne (nexibeo.com · completeaitraining.com), built together with Claude.

Measured 27 to 198 times cheaper, and more than twice as fast, than letting Claude's or Codex's own model do the clicking (same tasks, same loop; results).

Website: jevbrowsercontrol.com · Docs: jevbrowsercontrol.com/docs · License: MIT

What you can ask

Five everyday tasks, run live in Claude Code on September 26, 2026:

Costs are at the jevbrowsercontrol.com credits price (5× OpenRouter's); with your own OpenRouter key they are a fifth of that. Clicks that buy, pay, send, post or delete, and a Comment or Reply button that would publish what was typed, always stop and wait for your OK.

Install

Needs Node.js 18+ and Google Chrome. Nothing to install in the browser: the MCP server opens its own Chrome window.

1. Get a key. Create one in the dashboard (prepaid credits, from $10), or use your own OpenRouter key.

2. Add the MCP server with your key. Claude Code:

   claude mcp add -s user -e JBC_API_KEY=jbc_your_key jev-browser -- npx -y https://jevbrowsercontrol.com/downloads/jev-browser-control-mcp-0.4.0.tgz
   

   Codex:

   codex mcp add jev-browser --env JBC_API_KEY=jbc_your_key -- npx -y https://jevbrowsercontrol.com/downloads/jev-browser-control-mcp-0.4.0.tgz
   

   With your own OpenRouter key, use OPENROUTER_API_KEY=sk-or-v1-... instead. To keep the key out of Claude's and Codex's config, leave out -e/--env and put the same line in ~/.jev-browser-control/config.env.

3. Restart Claude Code or Codex and ask. A Chrome window opens on first use. When a site needs a login, sign in there yourself once; the window has its own profile (~/.jev-browser-control/chrome-profile) and keeps the logins.

From a clone, node scripts/install-agents.mjs does all of this for Claude Code and Codex at once, and also installs a skill and a Claude Code subagent that know how to use the tools. Several Claude Code and Codex sessions share the one window; when the session that opened it ends, the next one takes over.

What you get

• An MCP server for Claude Code, Codex and Claude Desktop, with 20 tools: browser_navigate, browser_snapshot, browser_click, browser_type, browser_select, browser_hover (menus that open on hover, like reaction pickers), browser_read, browser_screenshot, browser_clipboard (what a "Copy link" menu copied), browser_record (a screen recording of the browser, saved as MP4) and more, plus jev_task (hand a whole sub-task to Jev), jev_find and jev_check. By default it drives its own Chrome window through Playwright (browser mode).
• A Chrome extension (extension mode) for driving your everyday Chrome, with your normal profile and open tabs. It reads the page the same way and acts with trusted input. Run tasks from its side panel, or let Claude drive it.
• Your choice of who pays for Jev: your own OpenRouter key (free, this repo), prepaid credits from jevbrowsercontrol.com (one key, 5× OpenRouter's price), or any compatible endpoint such as TypeSafe direct.

Two editions, one codebase. This repo is the open-source edition: load extension/ and use your own OpenRouter key, credits, or a custom endpoint. The download on jevbrowsercontrol.com is the service edition, built from the same code with lib/edition.js set to service: it runs on credits only. npm run zip builds both (dist/ and dist/oss/).

Extension mode: your everyday Chrome

1. Install the extension. Clone this repo (or download the zip), open chrome://extensions, switch on Developer mode, click Load unpacked and pick the extension/ folder.

2. Choose a provider in the settings page that opens: paste an OpenRouter key or a jbc_ credits key, and press Test connection.

3. Connect Claude in extension mode:

   claude mcp add -s user -e JBC_MODE=extension jev-browser -- node /path/to/jev-browser-control/mcp/server.mjs
   

   or node scripts/install-agents.mjs --extension. Claude Desktop: add {"mcpServers": {"jev-browser": {"command": "node", "args": ["/path/to/mcp/server.mjs"], "env": {"JBC_MODE": "extension"}}}} to its config.

Grok Bot, ChatGPT, claude.ai and other cloud apps. They can't start a local program, so jevbrowsercontrol.com offers the same tools as a remote MCP server at https://jevbrowsercontrol.com/mcp (bearer: your jbc_ key). Switch on Remote AI apps in the extension's settings and the extension keeps an outbound connection to the relay; it's off by default, and clicks that buy, pay, send, post or delete always wait for your OK in the side panel. For Grok Bot there is a ready-made template: Jev Browser Operator.

Claude Code and OpenAI Codex agent files. agents/ has a skill (works in both) and a Claude Code subagent. node scripts/install-agents.mjs registers the MCP server and installs them for whichever of the two you have.

How a step works

page ──► numbered elements ──► one Jev call ──► stop gates ──► trusted input ──► page
         (extension, ~20 ms)   (~0.5 s)         (code)          (CDP, ~50 ms)

1. Snapshot. Visible buttons, links and fields become rows [n] role "label" value, with checked/selected state and nearby card text. Password and file fields are never listed. Open shadow roots are included.
2. One request, three answers. Jev is asked for the operation (CLICK, TYPE_TEXT, SELECT, PRESS_ENTER, SCROLL_*, BACK, WAIT, DONE, BLOCKED), speculatively for the best target of every operation, and, independently, whether the goal is already met. Code uses only the target that matches the chosen operation.
3. Stop gates in code. DONE counts only when the independent goal check agrees; clicks that buy, pay, send, post or delete stop for confirmation, and so does a Comment, Reply or Send button next to a text box that holds text (it would publish it); three actions without a visible change end the run; limits on actions, seconds and dollars.
4. Act. Before input, the element is re-checked (same node, same state, visible, not covered). If the page moved while Jev decided, nothing is clicked and it looks again. Only TYPE_TEXT needs words: a small LLM writes them from the goal and never invents personal data (it returns null, and the task stops with needs_input).

Jev's answer is always an element number that code maps back to a node it tagged itself. It never becomes a selector, coordinates or code.

The operation-plus-speculative-target design, the freshness guards and the page snapshot are adapted from Browser Use's Jev Ultrafast (MIT); see NOTICE.

Measured

Live runs on September 19, 2026 with typesafe/jev-1.13 (cost at OpenRouter prices; credits cost 5×):

Recorded traces are in results/.

Jev vs. Claude and Codex models

scripts/benchmark.mjs runs three live tasks (search Wikipedia and open an article; fill 7 fields of httpbin's order form without submitting; open the top Hacker News story's comments) through the same loop and page snapshot, changing only who answers each step's questions. With --session, the models run the way Claude Code and Codex run them: tool definitions in the system prompt, the whole conversation resent at every step, reasoning at medium effort, prompt caching on. Every model passed all three tasks; success is checked in code on the final page.

September 19, 2026, OpenRouter prices. Without --session (one bare call per step, reasoning off or low) the models cost about the same or less: $0.167, $0.247, $0.538 and $0.941. Prompt caching keeps the resent conversation cheap on short tasks like these; a long real session, which carries all its earlier work into every step, costs more. On Hacker News Jev reached the right page each time but its own goal check was unsure, so it ended as stuck or done_unconfirmed rather than done. Raw data: results/benchmark-session-2026-09-19.json and results/benchmark-2026-09-19.json.

Repository

npm install && npm install --prefix mcp   # playwright-core
npm test                    # 25 offline tests
OPENROUTER_API_KEY=... npm run e2e:browser            # browser mode, live, about $0.006
CHROME_PATH=... OPENROUTER_API_KEY=... npm run e2e    # extension mode, live, about $0.005
npm run zip                 # dist/ extension zips + MCP tarball

For npm run e2e, CHROME_PATH must be Chromium or Chrome for Testing: branded Chrome 137+ ignores --load-extension. Browser mode works with branded Chrome.

MCP server settings

In browser mode these can go in ~/.jev-browser-control/config.env (one KEY=value per line); the environment wins.

Safety and limits

• Remote control is off until you switch it on. While it's on, anyone with that jbc_ key can reach your browser through the relay, so keep the key secret and revoke it in the dashboard if it leaks. Remote callers can never skip the confirmation for irreversible clicks.
• The bridge listens on 127.0.0.1 only. Web pages can't connect (they can't forge a chrome-extension:// origin); other MCP sessions need a token from your home folder.
• Browser mode uses its own profile, so it only has the logins you make in its window. Extension mode acts in your normal profile, with all your logins: use a separate Chrome profile for risky work. In both, list sites like your bank as blocked sites.
• Page text is sent to Jev as untrusted data, but prompt injection can still mislead a model. Keep limits on and check results; Jev can pick a confident near-miss between look-alike names.
• Not supported yet: cross-origin iframes, canvas apps, file uploads, CAPTCHAs, and hover-only menus. English pages work best.

Credits

Created by Jeroen Erne (nexibeo.com · completeaitraining.com), built together with Claude. Loop design from Jev Ultrafast by Browser Use. Jev is a model by TypeSafe. This project is not affiliated with TypeSafe, Browser Use or Anthropic.