skills.rest
yethikrishna
yethikrishnaCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

agent-browser

Automate browser navigation, DOM inspection, form filling, clicks, downloads, screenshots, and data extraction for programmatic web tasks.

Updated Apr 16, 2026
Search Tags:#screenshots#session-management#scraping#browser-automation#visual-testing#form-filling
One-click install
npx skills add https://github.com/yethikrishna/humble --skill agent-browser-yethikrishna
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/yethikrishna/humble/tree/main/core/kortix-master/opencode/skills/KORTIX-system/agent-browser
Command: npx skills add https://github.com/yethikrishna/humble --skill agent-browser-yethikrishna

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Agent-browser removes manual, error-prone web interactions by letting AI agents navigate pages, discover and interact with UI elements, capture visual evidence, and extract structured content without hand-crafting selectors for every target page. It reduces context bloat for LLMs by using compact snapshot refs and enforces a consistent capture convention so results are organized and reproducible.

Core Features & Use Cases

  • Ref-based Interaction: Snapshot-driven refs (e.g., @e1) let agents click, fill, select, and type without sending full DOM blobs to the model.
  • End-to-end Workflows: Navigate, authenticate, persist session state, run interactions, and verify changes with snapshot diffs and visual screenshots.
  • Capture & Testing: Produce screenshots, PDFs, recordings, visual diffs, and performance profiles for debugging, regression testing, or documentation.
  • Security & Policy: Domain allowlists, action policies, content boundaries, and optional encryption for saved state mitigate unsafe or unexpected actions.
  • Use Cases: Automate login flows and authenticated scraping, run E2E tests and visual regressions, capture pages for archival, or generate evidence for audits and support tickets.

Quick Start

Open a page, snapshot interactive refs, perform interactions using those refs, and save screenshots to the ./screenshots/ directory.

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 navigation and form filling with AI agents?▼

Automate browser navigation and form filling by using snapshot-driven refs to interact with DOM elements without sending full DOM blobs to the model. Agents click, fill, select, and type using compact refs, reducing context bloat and enabling reproducible programmatic web tasks.

How does ref-based interaction work for web scraping and visual testing?▼

Ref-based interaction for web scraping and visual testing works by capturing snapshots of interactive elements and assigning them compact refs like @e1. Agents use these refs to perform actions and extract structured content, avoiding the need to hand-craft selectors for every target page.

Can I run parallel browser sessions and persist authenticated login state?▼

Yes, you can run parallel browser sessions and persist authenticated login state. The system supports session isolation, state persistence, and optional encryption for saved state, allowing agents to automate login flows and maintain authenticated sessions across multiple interactions.

Does browser automation support iOS simulators and local Chromium endpoints?▼

Browser automation supports iOS simulators through optional Appium integration and connects to a local Chromium or CDP endpoint. This allows agents to execute programmatic web tasks across desktop environments and mobile simulators seamlessly.

What is the best way to capture screenshots and visual diffs for end-to-end testing?▼

The best way to capture screenshots and visual diffs for end-to-end testing is to use the enforced capture convention that saves screenshots to a dedicated directory. You can also produce PDFs, recordings, and performance profiles for regression testing and debugging.

How do domain allowlists and action policies secure automated web scraping?▼

Domain allowlists and action policies secure automated web scraping by defining content boundaries and restricting unsafe or unexpected actions. This mitigates risks during programmatic navigation and data extraction by enforcing strict security rules.

Related Skills

Explore More Software Engineering Skills→
lemiglemig

setup-browser-cookies

Import Chromium browser cookies into a headless session for authenticated QA testing.

Community
Intermediate
winstonkoh87winstonkoh87
558

visual-verify-ui

Capture multi-viewport screenshots and compare DOM states for visual regression testing.

Community
Intermediate
vasilyu1983vasilyu1983
73

qa-testing-ios

Automate iOS app builds, tests, and UI automation with Xcode simulators.

Community
Advanced
harshaldhadukharshaldhaduk
1

browser

Automate web browser tasks using accessibility tree snapshots with element references.

Community
Intermediate