agent-browser

Automate browser interactions via the agent-browser CLI using ref-based selectors.

Updated Aug 23, 2026

Search Tags:#automation #CLI #browser #headless #web-scraping #form-filling #agent-browser

Ready to run

npx skills add https://github.com/jjeremycai/claudekit --skill agent-browser-jjeremycai

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/jjeremycai/claudekit/tree/main/plugins/compound-engineering/skills/agent-browser
Command: npx skills add https://github.com/jjeremycai/claudekit --skill agent-browser-jjeremycai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill automates browser interactions for AI agents, enabling reliable web tasks that previously required manual control.

Core Features & Use Cases

Headless browser automation using the agent-browser CLI.
Form filling, button clicks, navigation, and element interaction via ref-based selectors.
Web data extraction and screenshots for verification and reporting.
Use Case: Imagine an AI agent that signs in to a site, searches for products, captures details, and saves results to a report.

Quick Start

Open a URL and interact with elements using refs: agent-browser open https://example.com agent-browser snapshot -i agent-browser click @e1 agent-browser fill @e1 "Sample text"

Frequently Asked Questions about agent-browser

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I automate browser interactions for an AI agent without manual control?▼

Browser automation for AI agents is achieved by installing the agent-browser CLI via npm and executing Bash commands. This enables deterministic web tasks including visiting pages, filling forms, clicking controls, and taking screenshots across various websites.

How do I fill forms and click elements using CLI browser automation?▼

To fill forms and click elements, open a URL with agent-browser, take a snapshot to identify interactive elements, then use ref-based selectors. You can execute commands like 'agent-browser fill @e1 "Sample text"' and 'agent-browser click @e1' to interact with the page.

Does agent-browser require any specific dependencies or environments to run?▼

Yes, agent-browser requires installing the agent-browser CLI package via npm. It relies on Bash commands to execute deterministic automation tasks and uses ref-based selectors for element interaction within a headless browser environment.

What is the best way to scrape web data and capture screenshots for verification?▼

The best way to scrape web data and capture screenshots is using the agent-browser CLI. It supports web data extraction and taking screenshots directly through Bash commands, enabling AI agents to verify interactions and generate reports from various websites.

Can I use ref-based selectors for element selection in headless browser automation?▼

Yes, ref-based selectors are supported for element selection in headless browser automation. The agent-browser CLI uses refs like @e1 to target interactive elements, allowing you to click controls and fill forms deterministically via Bash commands.