browser-use/browsercode — explained in plain English
Analysis updated 2026-05-18
Automate browser tasks from a plain-English instruction
Scrape structured data like flight listings from a website
Test a local website and open a pull request with fixes
Save and reuse browser automation scripts across sessions
| browser-use/browsercode | ardupilot/node-mavlink | gonemedia/aipointer | |
|---|---|---|---|
| Stars | 97 | 96 | 98 |
| Language | TypeScript | TypeScript | TypeScript |
| Last pushed | — | 2025-08-26 | — |
| Maintenance | — | Quiet | — |
| Setup difficulty | moderate | moderate | easy |
| Complexity | 3/5 | 3/5 | 2/5 |
| Audience | developer | developer | general |
Figures from each repo's GitHub metadata at analysis time.
Requires your own AI model provider API key.
BrowserCode is an AI agent that can control a web browser on your behalf by writing and running JavaScript code. Instead of clicking and navigating with a simulated mouse and keyboard (which is how most browser automation works), it takes a more powerful approach: it writes JavaScript programs that talk directly to Chrome using the Chrome DevTools Protocol (CDP), a low-level interface that gives deep programmatic control over the browser. You install it with a single command in your terminal. Once running, you can give it natural-language instructions like "find me all flight details from New York to San Francisco tomorrow on Google Flights" or "test this local website and fix any bugs you find, then open a pull request." The agent figures out how to accomplish those tasks by writing browser automation scripts, which it saves for reuse in future sessions. The agent can connect to your existing browser tabs, create a new separate browser profile to work in the background, or spin up a cloud-hosted browser with stealth capabilities and proxy support. It supports multiple AI model providers, you supply your own API key. The tool is built on top of two other open-source projects: OpenCode (a terminal-based AI coding agent) and Browser Harness (a browser automation library from the same team). The key addition is a single core capability called browser_execute, which runs JavaScript in the context of a live browser session and returns the results, screenshots, logs, and values, back to the AI model. Anonymous usage data is collected by default but can be disabled by setting an environment variable.
An AI agent that controls a web browser by writing JavaScript against Chrome's low-level protocol, following natural-language instructions instead of clicking around.
Mainly TypeScript. The stack also includes TypeScript, JavaScript, Chrome DevTools Protocol.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.