Agent Browser Automation

0•

Automates browser navigation, UI testing, and visual dev server verification using interactive element snapshots and CLI commands.

Agent Browser provides Claude Code with powerful headless and headed browser automation capabilities through a persistent background daemon. It allows agents to open web pages, capture interactive snapshots with reference handles, fill forms, execute JavaScript, capture annotated screenshots, and persist authentication state. It is especially useful for automatically verifying local dev server outputs (Next.js, Vite, Nuxt), running automated end-to-end tests, and extracting structured web data during development workflows.

Características Principales

01Automated dev server verification with full-page and vision-annotated screenshots
02JavaScript evaluation and DOM diffing for visual and structural change detection
03Form submission, checkbox toggling, dropdown selection, and keyboard event simulation
040 GitHub stars
05Interactive element snapshotting with dynamic ref identifiers (@e1, @e2) and semantic locators
06Authentication state persistence and multi-session browser management

Casos de Uso

01Automating multi-step user authentication flows and saving session state for testing
02Visually verifying local dev server endpoints and frontend rendering after making code changes
03Extracting structured DOM data and debugging frontend issues with annotated visual snapshots