agent-test:page-testing

Automate DFS click-all route testing with agent-browser and capture interaction state snapshots.

3|1|Updated Mar 26, 2026
One-click install
npx skills add https://github.com/visual-z/agent-test --skill agent-test-page-testing
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-test:page-testing
Source: https://github.com/visual-z/agent-test/tree/main/skills/page-testing
Command: npx skills add https://github.com/visual-z/agent-test --skill agent-test-page-testing

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a deterministic approach for testing a single route by a subagent, exercising every interactive element with a depth-first click-all algorithm and capturing state snapshots for bug detection.

Core Features & Use Cases

  • Per-route DFS testing that treats the page as a maze and visits all interactive elements
  • Isolated, independent subagent sessions using agent-browser to ensure no cross-test interference
  • Comprehensive snapshot workflow with decision points, backtracking, and error handling
  • Clear phase structure from login to final-state reporting enabling automated QA workflows

Quick Start

Run a DFS click-all test on the target route using an isolated agent-browser session and capture screenshots for each interaction.

Frequently Asked Questions about agent-test:page-testing

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

FAQPage Schema
How do I automate exhaustive route testing for single-page web applications?

Automated route testing uses a depth-first search algorithm to click every interactive element on a web page. This DFS approach treats the route like a maze, systematically exploring dialogs and dropdowns while backtracking to capture all interaction states for bug detection.

What is the best way to prevent cross-test interference during web UI automation?

To prevent cross-test interference in web UI automation, you can run isolated independent sessions for each subagent. This ensures that testing one route's interactions does not affect the state or outcome of another route's QA workflow.

Does deterministic route testing require access to the application source code?

Deterministic route testing does not require source code access. It operates by interacting with the live web application purely through the UI, explicitly forbidding source code reading to ensure tests reflect actual user behavior and visual states.

Can I use agent-browser in headless mode for QA automation of complex workflows?

Yes, you can use agent-browser in headless mode for QA automation of complex workflows. It supports handling multi-step wizards, dropdowns, and dialog interactions while following a fixed lifecycle from open to close with explicit state snapshots.

How do you handle backtracking when exploring all interactive elements on a web route?

Backtracking during web route exploration is handled by capturing state snapshots at decision points. When a dead end or completed interaction is reached, the algorithm reverts to previous states to ensure exhaustive coverage of all remaining interactive elements.