User repository on GitHub
Claude thinks, Jev reacts: an MCP server that hands browser tasks from Claude to TypeSafe's Jev (~100 ms per decision).
- Forks
- 2
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Agents & MCP
Jev Reflex
An 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.
- Stars
- 0
- Language
- Python
- License
- MIT
- Reading time
- 3 min
- Published by
- Community
- Author
- msalvalaggio
- Last verified
- Oct 2026
- Languages
- Python
- Capabilities
- Open sourceRequires API key
Highlights
- The browse(url, goal) MCP tool opens a background Chrome tab, reads a numbered table of visible controls, and loops read → Jev picks action → execute.
- Jev chooses an operation and target in one request: CLICK, TYPE_TEXT, SELECT, SCROLL, WAIT, DONE or BLOCKED, each with a probability.
- For TYPE_TEXT, Jev picks which provided value goes in the selected field; with no values, a configurable small text model infers it from the goal.
- Safety before input: the target is rechecked against the live page and must exist, be visible and not be covered, or the decision is discarded.
- Stops on DONE or BLOCKED, the same page returning a third time (looping), or max_seconds; Claude gets status, per-step probabilities and final page text.
Watch out
Grounded in the truncated repo README and GitHub metadata read on 2026-10-15; the excerpt ends mid-way through how a run finishes. The ~100 ms figure is TypeSafe's, cited by the README, and the project is built on the browser loop from browser-use/jev-ultrafast.
Questions about this entry
Short answers drawn from the linked project or page.
An MCP server that lets Claude hand a whole browser task to Jev. Claude stays in charge of planning the task and verifying the result, while Jev makes each click, type or select decision — in about 100 ms per decision, per TypeSafe's figure.
Browser tools for LLM agents usually do one round trip to a frontier model per step — seconds and thousands of tokens per click, 15 of them for a 15-click form. Most steps are choices from a list, which is what Jev is built for; the README calls this System 1 for Jev and System 2 for Claude.
Claude calls browse with a URL, a goal that says when to stop, and the exact strings to type. jev-reflex opens a background tab, reads a numbered table of every visible control, and loops: Jev picks an operation and target with a probability, the action is executed, repeat until DONE or BLOCKED.
CLICK, TYPE_TEXT, SELECT, SCROLL, WAIT, DONE or BLOCKED, each with a probability. For TYPE_TEXT, Jev also picks which of Claude's values belongs in the selected field; if no value fits, the run stops and Claude takes over.
Before any input, the target is rechecked against the live page: it must still exist, be visible and not be covered by another element. If the page changed, the decision is discarded and the loop observes again. There is also loop detection and a hard max_seconds limit.
Claude receives the status, every step with its probability, and the final page text. If the run did not finish, it also receives the element table so Claude can continue or take over.
From browser-use/jev-ultrafast. jev-reflex exposes it as an MCP tool and adds what an orchestrating LLM needs: a verifiable result, loop detection, a hard time limit and a handoff when Jev gets stuck.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Jev in front of the HeyGen MCP
X: Playwright CLI + Jev vs Playwright MCP
X: A local MCP server that cuts computer use from 23s to 0.8s. Cartwright built Cortex, a local MCP server where…
X: A side panel that drives any site's WebMCP tools
More Jev resources like this
Chrome MV3 extension plus a local MCP server exposing 50 browser tools; its chrome_act_toward_goal micro-loop uses TypeSafe Jev System One or a local decider to pick the next action on an 85%+ pruned, Shadow-DOM-pierced DOM tree, dispatching trusted CDP events in logged-in Chrome without stealing focus.
Jev Agents & MCP#community#mcp#browser-automation
Community
GoldenLoaf24h/browserpawTypeScript
GoldenLoaf24h/browserpawTypeScriptLets 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-desktopPythonA plugin-based agent harness (FastAPI + Vue 3) bundling a chat GUI, multi-provider model access, an MCP client and marketplace, skills, workflow studio, sandboxed execution and a Tauri desktop shell. It integrates TypeSafe AI's Jev for structured Choice, Score and Noul decisions.
Jev Agents & MCP#community#github#agents
Community
andyqiuqiubo/andy-harnessTypeScript
andyqiuqiubo/andy-harnessTypeScriptMCP server that lets Claude Code, Codex and other agents drive a real Chrome browser while JEV makes the small typed decisions — which element, which suggestion, did the click work — with calibrated confidence driving act, verify or ask.
Jev Agents & MCP#community#catalog-import#logicrw
Community
legostin/jev-mcpTypeScript
legostin/jev-mcpTypeScriptA 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.
Jev Agents & MCP#community#catalog-import#logicrw
Community
tongyun1/jev-in-the-loopPython
tongyun1/jev-in-the-loopPythonCodex 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-skillPythonFrom 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.