How does it handle elements that change dynamically?

The skill uses a snapshot mechanism that generates temporary unique refs (@e1, @e2, etc.) based on the accessibility tree, ensuring precise interaction even if CSS classes change.

Can I see the network requests made during the test?

Yes, the skill includes a network auditing step that records all XHR and Fetch requests, specifically flagging 4xx and 5xx errors in the final test report.

Does it support custom JavaScript execution?

Yes, if standard element selection fails, the skill can execute custom JavaScript via the 'eval' command to interact with complex or non-standard web components.

What dependencies are required for this skill?

This skill requires the agent-browser CLI (vercel-labs/agent-browser) to be installed on your system for Claude to control the browser instance.

Where are the testing reports stored?

Reports are automatically generated as Markdown files and stored in the 'testing-report/' directory within your project root, organized by timestamp.

Agent Browser Integration Testing

Name: Agent Browser Integration Testing
Author: sancodeee

bysancodeee

0•

ブラウザ自動化

Automates end-to-end browser integration testing with real-time reporting, network auditing, and screenshot evidence.

This skill empowers Claude to perform professional-grade web automation and E2E testing using the agent-browser CLI. By leveraging an accessibility tree snapshot mechanism, it ensures stable element positioning even on highly dynamic websites. The skill follows a strict standard operating procedure (SOP) that includes real-time report generation, network request auditing to catch API errors, and automated screenshot capturing for visual verification. It is ideal for developers and QA engineers who need reliable, documented testing workflows for complex business logic, login flows, and regression suites.

主な機能

01Atomic browser control using accessibility tree snapshots for stable element targeting

020 GitHub stars

03Advanced handling for complex UI components like custom dropdowns and dynamic modals

04Integrated network auditing to capture and analyze XHR/Fetch requests and status codes

05Real-time, standardized Markdown report generation with progressive logging

06Automated screenshot evidence capture with full-page and element-specific support

ユースケース

01Auditing web application network traffic to identify failing API endpoints or performance bottlenecks

02Automating multi-step functional testing for user login and checkout workflows

03Performing regression testing across different web environments with standardized reporting

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add sancodeee/sen-claude-hub agent-browser-integration-testing

For use in Claude.ai and ChatGPT

Download Skill