visual-development-loop

Verifies UI changes with browser screenshots, accessibility snapshots, and smoke tests.

3|Updated Apr 8, 2025
One-click install
npx skills add https://github.com/ZanzyTHEbar/dragonarchy --skill visual-development-loop-zanzythebar
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: visual-development-loop
Source: https://github.com/ZanzyTHEbar/dragonarchy/tree/main/packages/opencode/.config/opencode/skills/visual-development-loop
Command: npx skills add https://github.com/ZanzyTHEbar/dragonarchy --skill visual-development-loop-zanzythebar

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Frontend changes often ship without real browser evidence, leaving rendering bugs, console errors, and accessibility regressions undetected until users report them. ## Core Features & Use Cases - Repeatable Browser Verification: Drives Playwright MCP, browser-use MCP, or CDP harnesses to capture before-and-after screenshots and accessibility snapshots for any UI change. - Structured Proof Workflow: Produces falsifiable verdicts (VERIFIED, NOT VERIFIED, INCONCLUSIVE) with captured evidence for claims that a UI works. - Use Case: After restyling a checkout page, run the loop to capture baseline and post-change screenshots at desktop and mobile viewports, check console errors, and run the repo's smoke test before delivery. ## Quick Start Verify that the updated dashboard page renders correctly by capturing before and after screenshots and checking for console errors.

Frequently Asked Questions about visual-development-loop

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

FAQPage Schema
How do I verify a UI change with browser screenshots?

Start the local app, capture a baseline screenshot and accessibility snapshot, make one change at a time, then capture the after-state with the same viewport and route. Check console errors, failed network requests, and layout regressions before concluding.

What tools does this visual testing workflow use?

It prefers Playwright MCP for navigation, snapshots, and screenshots, uses browser-use MCP only when credentials exist, and falls back to CDP harnesses for Chromium or Electron debug-port work. Repo-native smoke tests provide the narrowest automated check.

When should I use CDP instead of Playwright MCP?

Use CDP harnesses when working with Chromium or Electron debug ports or when you need low-level runtime signals that Playwright MCP cannot expose. Playwright MCP remains the default for standard navigation and interaction tasks.

How are screenshots and artifacts handled for privacy?

Prefer inline summaries when screenshots, traces, or network bodies may contain private content. Save artifacts under /tmp/verify-this/<claim-slug>/ only when safe, name them by phase and viewport, and remove temporary browser profiles afterward.

What selectors should I use for browser automation?

Prefer stable selectors such as roles, labels, landmarks, and project-owned data-* attributes. Avoid coordinate clicks unless a fresh screenshot proves the target and no semantic selector exists.