dogfood_skill

Overview

This skill systematically explores and tests a web application to find bugs, UX issues, and other quality problems. It produces a structured, reproducible report for every finding, with step-by-step screenshots, repro videos when needed, and clear reproduction steps so issues can be handed directly to responsible teams.

How this skill works

I start a named browser session against the target URL, orient myself with an initial annotated snapshot, and then explore the app methodically: navigation, forms, workflows, and edge cases. When I find an issue I immediately document it with the appropriate evidence (annotated screenshots for static issues; recorded repro videos plus step screenshots for interactive issues) and append a numbered ISSUE block to the report. At the end I summarize totals and close the session.

When to use it

• When you want an exploratory QA pass of a site, app, or platform
• When you need reproducible evidence (screenshots/videos) for bugs
• Before a release to catch UX regressions and console errors
• For focused testing of a feature area (billing, signup, settings)
• To hand findings directly to engineering or product teams

Best practices

• Provide only the Target URL unless you need custom session name, scope, or auth
• Don’t ask clarifying questions unless login is required and credentials are missing
• Verify an issue is reproducible before recording evidence
• Match evidence to issue type: screenshots for static issues, videos for interactive flows
• Document each issue immediately with numbered repro steps that reference screenshots

Example use cases

• Dogfood vercel.com to find layout issues, JS errors, and broken links
• Exploratory QA of a staging build at http://localhost:3000 before deployment
• Focused audit of the billing page to validate form flows and error handling
• Regression check after frontend changes to verify no new console errors
• Hand-off ready bug reports (video + screenshots + steps) to engineering