sawyerhood/dev-browser

A Claude Skill to give your agent the ability to use a web browser

View on GitHub ↗Jump to charts ↓

Summary Information

Updated 5 minutes ago
Added to GitGenius on September 1st, 2026
Created on December 2nd, 2025
Open Issues & Pull Requests: 18 (+0)
GitHub issues: Enabled
Number of forks: 432
Total Stargazers: 6,593 (+0)
Total Subscribers: 19 (+0)

Charts & Analytics

Fetching additional details & charts...

Issue Activity (beta)

Issue API getrepoissuespagesummary failed: 429 Rate limit exceeded. Please try again later.

Detailed Description

Dev Browser is a browser automation tool that gives Claude Code agents and developers the ability to control web browsers through sandboxed JavaScript scripts.

The tool solves the problem of enabling AI agents to interact with web applications by providing a sandboxed execution environment where scripts run in a QuickJS WASM sandbox with no access to the host system. It uses Playwright's API to control browsers, supporting standard operations like navigation, clicking, form filling, locators, evaluation, and screenshots. The tool can either connect to an existing Chrome instance or automatically launch a fresh Chromium browser, maintaining persistent page state across multiple script invocations.

Developers should choose this tool if they need to automate browser interactions within Claude Code workflows or integrate web automation into AI agent systems. It suits projects where sandboxed execution is a requirement, as scripts cannot access the host filesystem or network. The tool is designed specifically as a Claude Code skill, making it most valuable for developers working within that ecosystem. For those concerned about resource usage, the tool includes configurable idle timeout functionality that automatically closes unused Chromium instances while preserving their profile directories, cookies, and login state for later reuse.

The project maintains an embedded skill definition that can be installed or refreshed across multiple agent targets, with the help output providing current API reference documentation for LLM consumption. Windows support is handled through native executable distribution during installation, with platform-specific connection instructions documented. The tool provides multiple configuration layers for idle timeout settings, allowing users to set defaults globally, per-environment, or per-invocation, with the effective setting visible through status commands. Pre-approval mechanisms exist for Claude Code integration to eliminate permission prompts on repeated use, with safety documented through the sandboxed execution model.