SkillsLib.ai

Agent Browser CLI

Token-efficient browser automation for AI agents

3.8(6 reviews)
10,000+ downloads
Updated Oct 2026
V
by vercel-labs
Technology & SoftwareData & AnalyticsAI/Prompt Engineer

What You Can Do

Automate Chrome/Chromium browser interactions using compact accessibility-tree snapshots that use 200-400 tokens instead of raw HTML. Navigate pages, click elements, fill forms, extract data, and take screenshots. Run on local Chrome or AWS Bedrock AgentCore cloud browsers with persistent profiles. Supports parallel sessions, MCP integration, and eve agent embedding.

Features

Snapshot-and-Ref Workflow

Get accessibility-tree snapshots with compact @eN refs instead of parsing raw HTML. Reduces token overhead by 50-75% compared to raw DOM inspection

Local & Cloud Browser Providers

Run on local Chrome/Chromium via CDP or use AWS Bedrock AgentCore for managed cloud sessions with automatic credential resolution

Element Interaction

Click, fill, type, select dropdowns, upload files, hover, drag-and-drop, and send keyboard input using snapshot refs or semantic locators

Data Extraction & Screenshots

Extract text, HTML, attributes, page titles, and URLs. Capture full-page or cropped screenshots programmatically

Form & Authentication Automation

Fill complex forms, handle multi-step login flows, maintain session state with persistent browser profiles and localStorage

Parallel Session Management

Run multiple independent browser sessions in parallel with isolated state, cookies, and profiles using named session IDs

Page Reading & Documentation Parsing

Read rendered DOM content, fetch documentation with markdown negotiation, filter by sections, and discover llms.txt endpoints

MCP & Agent Integration

Expose tools via Model Context Protocol server and @agent-browser/eve extension for seamless AI framework embedding

Example Output

Navigate a Page and Extract Data

code
agent-browser open https://example.com
agent-browser snapshot -i
agent-browser click @e3
agent-browser screenshot result.png

Opens a webpage, takes an interactive-element snapshot showing clickable refs, clicks an element, and captures a screenshot of the result.

Automate Form Submission

code
agent-browser fill @e2 "user@example.com"
agent-browser fill @e3 "password123"
agent-browser click @e5
agent-browser wait --load networkidle

Fills form fields with credentials, clicks submit, and waits for network idle to ensure the page has fully loaded.

Run on AWS Bedrock AgentCore

code
AGENTCORE_REGION=us-east-1 agent-browser -p agentcore open https://example.com
agent-browser snapshot -i
agent-browser click @e1
agent-browser close

Launches a managed cloud browser session on AWS, takes a snapshot, interacts with elements, and closes the session.

What's Included

  • agent-browser CLI: Fast command-line tool installed via npm, backed by Chrome DevTools Protocol for local or cloud browser control
  • Snapshot & Navigation Commands: snapshot, open, close, navigate, press, scroll, read with markdown parsing and section filtering
  • Interaction Commands: click, fill, type, select, check, upload, drag, hover, and element inspection tools (text, html, attr, value)
  • AWS Bedrock AgentCore Provider: Cloud browser provider with auto-credential resolution, persistent profiles, live-view monitoring, and multi-region support
  • Session & Parallel Execution: Named session isolation, multiple independent browsers, auto-restore, configurable idle timeouts, and worktree-scoped sessions
  • MCP & Eve Integration: Stdio MCP server for framework embedding, paginated tool discovery, and @agent-browser/eve extension for agent sandboxes

Who It's For

  • AI and ML Engineers
  • Automation and Bot Developers
  • QA and Test Automation Specialists
  • Web Data and API Engineers
  • Claude Code and Agent Builders

Best For

  • AI agent web task automation
  • Web scraping with JavaScript execution
  • Automated form filling and submission
  • Functional and E2E testing
  • Multi-step browser workflow automation

You might also like

FREE
AI/Prompt Engineer3.5(4)
Find Skills

Help users discover specialized agent skills by searching the open skills ecosystem when they ask 'how do I do X' or want to extend capabilities. Verify skill quality based on install counts and source reputation, then install skills directly for users. The skill guides users through understanding their needs, finding the best matches, and enabling immediate access to new agent capabilities.

V
by vercel-labs
FREE
AI/Prompt Engineer3.8(5)
Algorithmic Art Generator

This skill enables you to create computational art through a two-phase process: first develop an algorithmic philosophy or manifesto for your art movement, then express it in executable p5.js code. You can generate interactive, seeded generative art with full parameter exploration, create particle systems and flow fields, and develop gallery-quality algorithmic compositions with reproducible randomness.

A
by anthropics
FREE
Software Engineer3.5(6)
Vercel Agent Skills

Access Vercel's official collection of agent skills for building and deploying modern web applications. Optimize React component architecture using composition patterns to avoid boolean prop proliferation and build scalable component libraries. Deploy applications to Vercel with flexible workflows supporting git integration, CLI authentication, and multiple fallback methods. Apply React and Next.js performance guidelines from Vercel Engineering to write efficient code and improve Core Web Vitals.

V
by vercel-labs
Free