midscene-testing

Automate headless browser testing with AI-guided prompts and merge HTML reports.

4|Updated Jan 29, 2026

Search Tags:#puppeteer #ui-testing #browser-testing #ai-guided #midscene #report-merge

Ready to run

npx skills add https://github.com/danmestas/wardrobe --skill midscene-testing

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: midscene-testing
Source: https://github.com/danmestas/wardrobe/tree/main/skills/midscene-testing
Command: npx skills add https://github.com/danmestas/wardrobe --skill midscene-testing

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates ad-hoc browser testing and UI validation using AI-guided prompts.

Core Features & Use Cases

AI-guided, headless browser testing to validate multi-step frontend workflows.
Per-command screenshots with vision-model feedback to guide interaction decisions.
Consolidation of per-step HTML reports into a single navigable document for rapid review.

Quick Start

Run Midscene to smoke-test a web app by connecting to it, issuing natural-language prompts to drive UI actions, and merge the resulting HTML reports into a single navigable document.

Frequently Asked Questions about midscene-testing

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

FAQPage Schema

What is AI-guided browser smoke testing and how does it work?▼

AI-guided browser smoke testing uses natural-language prompts to drive headless Puppeteer interactions, capturing per-command screenshots with vision-model feedback to validate multi-step frontend workflows dynamically.

How do I automate UI validation for multi-step frontend workflows?▼

Automate UI validation by issuing natural-language prompts to a headless Puppeteer instance, which executes browser actions, captures screenshots per command, and consolidates per-step HTML reports for review.

Can I use Puppeteer for natural-language interactions in browser testing?▼

Yes, Puppeteer runs headless browser sessions where AI-guided natural-language prompts drive UI actions, capturing screenshots and validating dynamic frontend behavior without manual scripting.

How do I consolidate per-step HTML reports into a single navigable document?▼

Consolidate per-step HTML reports by running the merge-reports.mjs pipeline with step labeling support, merging individual test artifacts into a single navigable document for rapid review.

Does this approach work for validating dynamic frontend behavior?▼

Yes, the headless Puppeteer workflow applies AI-guided prompts to validate dynamic frontend behavior, using per-command screenshots and vision-model feedback to guide interaction decisions during multi-step workflows.

What are the limitations of AI-guided smoke testing for web apps?▼

AI-guided smoke testing relies on headless Puppeteer and vision-model feedback, meaning it validates frontend workflows but does not execute unit tests or backend API validation outside the browser context.