local-testing

Automate local end-to-end testing of web, Electron, and macOS desktop bots.

1|Updated May 9, 2026
One-click install
npx skills add https://github.com/duwenji/generative-ai-oss-tutorials --skill local-testing-duwenji
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: local-testing
Source: https://github.com/duwenji/generative-ai-oss-tutorials/tree/main/sandbox/lobe-chat/.agents/skills/local-testing
Command: npx skills add https://github.com/duwenji/generative-ai-oss-tutorials --skill local-testing-duwenji

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires agent-browser, node, swift, ffmpeg, curl, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill reduces the friction of validating LLM/agent behavior by automating repeatable local testing for both web/Electron UIs and native macOS desktop bots.

Core Features & Use Cases

  • agent-browser (Electron/Web UI testing): Drives Chromium-based apps via CDP to navigate, snapshot UI elements, interact with DOM controls, and verify results with screenshots and diffs.
  • osascript (native macOS app bot testing): Automates real desktop apps (WeChat, Discord, Telegram, Slack, Lark/飞书, QQ) through AppleScript + Accessibility to send bot messages and capture visual proof.
  • End-to-end test harness for local dev: Includes helper scripts and references for setting up Electron dev with CDP, recording demos, and injecting authenticated sessions for local LobeHub testing.

Use case: You have a local LobeHub Electron app and want to confirm that a bot responds correctly—first automate UI interactions in Electron via agent-browser, then run desktop app bot checks via osascript, and finally capture screenshots/videos for comparison.

Quick Start

Instruct the AI to prepare a local test by starting Electron dev with CDP, then using agent-browser to open your target URL, snapshot interactive elements, send a message, and capture a verification screenshot.

Frequently Asked Questions about local-testing

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

FAQPage Schema
How do I automate local testing for LLM bots in native macOS chat apps?

You can automate local testing for native macOS bots using osascript, which leverages AppleScript Accessibility to send messages and capture visual proof in apps like WeChat, Slack, and Telegram.

Can I use CDP automation to test an Electron web UI locally?

Yes, CDP automation supports testing Electron web UIs locally by driving Chromium-based apps via agent-browser to navigate, snapshot DOM elements, interact with controls, and capture verification screenshots.

What's the best way to verify LLM agent behavior with visual screenshots?

Visual verification of LLM agent behavior is handled by capturing screenshots and diffing UI elements through agent-browser for web apps or osascript for native macOS desktop applications.

Does agent-browser support injecting authenticated sessions for local LobeHub testing?

Yes, the local test harness includes helper scripts and references for injecting authenticated sessions and setting up Electron dev with CDP specifically for local LobeHub testing.

Do I need Swift and ffmpeg to run local end-to-end bot tests on macOS?

Swift and ffmpeg are required dependencies for running local end-to-end bot tests, supporting osascript automation for native apps and recording demos or capturing visual proof.