skills.rest
t0lab
t0labOFFICIAL
Explore all skills by this author →
WebsiteGitHub

agent-browser

Automate browser navigation, form filling, clicking, and screenshots via CDP.

2|Updated Apr 10, 2026
Search Tags:#cdp#sessions#e2e-testing#screenshot#web-scraping#browser-automation#form-filling
One-click install
npx skills add https://github.com/t0lab/harness-kit --skill agent-browser-t0lab
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/t0lab/harness-kit/tree/main/.agents/skills/agent-browser
Command: npx skills add https://github.com/t0lab/harness-kit --skill agent-browser-t0lab

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill provides a dependable, scriptable CLI interface that enables AI agents to perform deterministic browser automation—navigating pages, filling forms, clicking elements, taking screenshots, extracting content, and running E2E checks—so users avoid manual, error-prone DOM scraping and heavy Playwright/Puppeteer dependencies.

Core Features & Use Cases

  • Ref-based Interaction: Compact snapshots produce stable @refs for precise element actions without sending full DOM content to the model.
  • Session & Auth Management: Save/load authenticated state, persistent profiles, and an auth vault to manage credentials securely for repeatable workflows.
  • Capture, Profiling & Integration: Full-page screenshots, PDF export, video recording, performance profiling, proxy support, Electron and Slack automation, and cloud/browser-provider integration for QA and exploratory testing.

Quick Start

Install and load the agent-browser skill, open the target URL, snapshot to discover refs, and use ref-based commands to interact and capture evidence.

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 AI agents without sending full DOM content to the model?▼

Browser automation for AI agents uses compact snapshots to produce stable @refs for precise element actions. This approach enables navigation, form filling, and clicking without sending full DOM content to the model or relying on heavy Playwright dependencies.

How do I manage authenticated sessions for repeatable web scraping workflows?▼

Authenticated sessions for web scraping workflows are managed through persistent profiles and an auth vault. This functionality saves and loads authenticated state, allowing you to execute repeatable scraping and form-filling tasks without repeatedly entering credentials.

Does browser automation via CDP require a specific browser environment setup?▼

Browser automation via CDP requires a Chrome or Chromium target accessible through the Chrome DevTools Protocol. You also need the agent-browser CLI with skills preloaded to execute commands for navigation, profiling, and state management.

Can I capture screenshots and record video during exploratory QA testing?▼

Screenshots and video recording are fully supported during exploratory QA testing. You can capture full-page screenshots, export PDFs, record video, and run performance profiling to gather evidence and analyze web application behavior.

What is the best way to automate Electron and Slack applications for E2E testing?▼

Automating Electron and Slack applications for E2E testing is handled directly through the CLI interface. It provides scriptable commands to interact with these platforms, enabling deterministic workflow automation without manual DOM scraping.

What are the limitations of using ref-based interaction for web scraping?▼

Ref-based interaction for web scraping relies on compact snapshots to generate stable @refs, meaning dynamic content changes may require re-snapshotting. It is designed for deterministic workflows, so highly volatile single-page applications might need repeated state management.

Related Skills

Explore More Software Engineering Skills→
npc-livenpc-live
168

agent-browser

Automate browser interactions via Chrome/CDP using the agent-browser CLI.

Official
Intermediate
antriksh-29antriksh-29

qa-only

Generate structured QA reports with health scores, screenshots, and reproducible steps.

Community
Advanced
LuizEduPPLuizEduPP

agent-browser

Automate browser navigation, form filling, and page interaction via CLI.

Community
Advanced
HouseGarofaloHouseGarofalo
3

browserless

Automate headless Chrome browser tasks via REST API and WebSocket.

Official
Intermediate