Claude Skill

browser-verification

Verify frontend changes in a real browser, including layout, console errors, responsive states, network failures, screenshots, and interactions. Use after local or deployed web UI changes, visual fixes, forms, navigation, asset updates, canvas or 3D work, or whenever browser veri

LLM Mart · 0 points · 2 views 0 listing impressions 0 install-command copies
Download markoblogo-abvx-agent-skills-skills_browser-verification-37a7146.zip · 2 KB

Install

skills CLI npx skills add https://github.com/markoblogo/abvx-agent-skills/tree/main/skills/browser-verification
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install markoblogo-abvx-agent-skills@llmmart
Git git clone https://github.com/markoblogo/abvx-agent-skills.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole markoblogo/abvx-agent-skills collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Browser Verification

Trust the browser, not static guesses. Use real rendering for frontend claims.

Setup

Prefer the repo's existing Playwright setup. If none exists, install browser tooling in a local tool directory or temporary environment rather than adding a production dependency.

Use a local static or dev server when file URLs would not match production behavior.

For security review of live targets, request routing through authorized-security-router first so scope, allowed actions, and non-destructive boundaries are explicit.

Verification Loop

  1. Start the app or static server.
  2. Open the target URL in a browser-controlled session.
  3. Check console errors and failed network requests.
  4. Verify key selectors render and have nonzero bounding boxes.
  5. Verify images/assets are complete and natural dimensions are nonzero.
  6. Check desktop and mobile viewports.
  7. Check horizontal overflow: document.documentElement.scrollWidth <= innerWidth + 1.
  8. Exercise the primary interaction path.
  9. Capture screenshots only when useful for review or debugging.
  10. Stop local servers before finalizing.

Common Assertions

  • page loaded with expected title or heading;
  • no failed requests for CSS, JS, images, fonts, or data;
  • no severe console errors;
  • target controls are visible and enabled;
  • form submission or navigation reaches expected state;
  • responsive layout has no text overlap or horizontal scroll;
  • animated/canvas/3D surfaces have nonblank rendered pixels when relevant.

Guardrails

  • Do not commit local Playwright tooling unless the repo already owns browser tests.
  • Do not claim visual verification if only static checks ran.
  • If browser install fails, report the fallback checks honestly.
  • For authenticated or paid flows, avoid real side effects unless explicitly approved.

Final Report

Include URLs, viewports, checks run, commands, failures found, and whether screenshots were captured.

Files (abvx-agent-skills)
  • agents
    • openai.yaml 200 B
      Loading…
  • SKILL.md 2.4 KB
    ---
    name: browser-verification
    description: Verify frontend changes in a real browser, including layout, console errors, responsive states, network failures, screenshots, and interactions. Use after local or deployed web UI changes, visual fixes, forms, navigation, asset updates, canvas or 3D work, or whenever browser verification should replace static guessing.
    license: MIT
    metadata:
      abvx_status: experimental
      abvx_origin: adapted
      abvx_eval_tier: fixture_checked
    ---
    
    # Browser Verification
    
    Trust the browser, not static guesses. Use real rendering for frontend claims.
    
    ## Setup
    
    Prefer the repo's existing Playwright setup. If none exists, install browser tooling in a local tool directory or temporary environment rather than adding a production dependency.
    
    Use a local static or dev server when file URLs would not match production behavior.
    
    For security review of live targets, request routing through `authorized-security-router` first so scope, allowed actions, and non-destructive boundaries are explicit.
    
    ## Verification Loop
    
    1. Start the app or static server.
    2. Open the target URL in a browser-controlled session.
    3. Check console errors and failed network requests.
    4. Verify key selectors render and have nonzero bounding boxes.
    5. Verify images/assets are complete and natural dimensions are nonzero.
    6. Check desktop and mobile viewports.
    7. Check horizontal overflow: `document.documentElement.scrollWidth <= innerWidth + 1`.
    8. Exercise the primary interaction path.
    9. Capture screenshots only when useful for review or debugging.
    10. Stop local servers before finalizing.
    
    ## Common Assertions
    
    - page loaded with expected title or heading;
    - no failed requests for CSS, JS, images, fonts, or data;
    - no severe console errors;
    - target controls are visible and enabled;
    - form submission or navigation reaches expected state;
    - responsive layout has no text overlap or horizontal scroll;
    - animated/canvas/3D surfaces have nonblank rendered pixels when relevant.
    
    ## Guardrails
    
    - Do not commit local Playwright tooling unless the repo already owns browser tests.
    - Do not claim visual verification if only static checks ran.
    - If browser install fails, report the fallback checks honestly.
    - For authenticated or paid flows, avoid real side effects unless explicitly approved.
    
    ## Final Report
    
    Include URLs, viewports, checks run, commands, failures found, and whether screenshots were captured.
    
  • SKILL_CARD.md 1.3 KB
    Loading…

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.