browser-harness-skill

Automate real Chrome browser interactions via raw CDP with screenshots and JavaScript evaluation.

356|66|Updated Oct 13, 2025

Search Tags:#javascript #browser automation #form filling #cdp #chrome #screenshot #tab management

Ready to run

npx skills add https://github.com/zeenie-ai/OpenCompany --skill browser-harness-skill

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: browser-harness-skill
Source: https://github.com/zeenie-ai/OpenCompany/tree/main/server/skills/web_agent/browser-harness-skill
Command: npx skills add https://github.com/zeenie-ai/OpenCompany --skill browser-harness-skill

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill lets you automate and inspect a user's real Chrome browser when traditional browser automation is too limited, especially for logged-in sessions, bot-protected sites, shadow DOM, canvas-heavy interfaces, and interactions that depend on visual layout.

Core Features & Use Cases

Vision-first browser control: Capture screenshots, click precise coordinates, type into fields, scroll pages, and verify results visually.
DOM and tab inspection: Use JavaScript to read page content, inspect attributes, target iframes, and manage tabs without relying on accessibility trees.
Use cases: Complete account-specific workflows in the user's existing browser session, interact with sites that resist automation, or perform UI actions that require exact on-screen positioning.

Quick Start

Ask the assistant to open the target page in your real Chrome, inspect it with a screenshot, and complete the next interaction using the browser harness helpers.

Frequently Asked Questions about browser-harness-skill

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

FAQPage Schema

How do I automate form filling on logged-in Chrome sessions with bot protection?▼

Automate logged-in Chrome sessions by using raw CDP to capture screenshots, click precise coordinates, and fill forms visually. This bypasses bot protection and structured automation limitations by interacting with the real browser session directly.

What is vision-first browser control and when do I need it for web automation?▼

Vision-first browser control captures screenshots to guide coordinate clicks and visual verification. You need it for canvas-heavy interfaces, shadow DOM, and iframe-heavy pages where structured browser tools cannot express the required interactions.

How do I inspect shadow DOM and iframe content in Chrome using JavaScript evaluation?▼

Inspect shadow DOM and iframe content by evaluating JavaScript directly in the real Chrome browser via CDP. Read page content, inspect attributes, and target iframes without relying on accessibility trees for deterministic navigation.

Can I manage multiple browser tabs in Chrome through raw CDP automation?▼

Yes, manage multiple Chrome browser tabs through raw CDP automation. Open target pages, switch tabs, and execute interactions across different tabs within your existing logged-in browser session.

Does coordinate-based clicking work for canvas-heavy interfaces where structured automation fails?▼

Yes, coordinate-based clicking works for canvas-heavy interfaces where structured automation fails. Capture screenshots and click exact on-screen positions to perform visual workflows that depend on layout rather than DOM structure.