Credentials. The agent drives your existing Chrome profile, so owned tabs share your logins; use a separate profile for experiments.
Who it is forA developer evaluating browser agents who wants to see why one is fast before adopting it.
The problemMost browser agents are slow because they screenshot and call a big model for every step, and you cannot see why they chose an action.
The outcomeA local demo that runs a real task and shows numbered elements, operation and target probabilities and every executed action.
What you need first
- Python with uv
- Chrome with remote debugging allowed (connected through Browser Harness, installed by uv sync)
- TYPESAFE_API_KEY and TEXT_MODEL_API_KEY (an OpenRouter key in the example config)
- Live examples make paid API calls
Steps
- Run
git clone https://github.com/browser-use/jev-ultrafast.gitandcd jev-ultrafast - Run
uv syncthencp .env.example .envand add TYPESAFE_API_KEY and TEXT_MODEL_API_KEY - Start the inspector with
uv run jevand open http://127.0.0.1:8766 - Click Start demo, then Run automatically; use Choose next to pause before each action
- If Chrome does not connect, run
uv run browser-harness --doctorand allow remote debugging
Prompt for your agent
Set up browser-use/jev-ultrafast locally, start the inspector with uv run jev, run the demo step by step with Choose next, and summarize for each step the operation chosen, its probability and the target element.
Caveat.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for browser-use/jev-ultrafast and "Run a 7 second browser agent and watch every decision"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer