Native Browser AI Skill

Accessibility and agent-readiness checks run locally in browser with no server-side inference egress.

Runs in browserPrivacy-preservingNo model egress

This demo represents a reusable browser-native skill pattern for enterprise teams that need fast, private page-readiness checks before agent automation is enabled.

Page Markup Input

Try it yourself in Chrome

Copy this prompt and paste it into your Chrome Gemini sidebar to run an accessibility audit on any webpage — no API key, no cloud, fully private.

AI Skill Manifest

{
  "name": "Gemini Nano Accessibility Auditor",
  "version": "1.0",
  "description": "WCAG 2.2 accessibility audit using on-device Gemini Nano — no server, no API key",
  "prompt": "You are an expert accessibility auditor. Analyze the current webpage and provide:\n1. WCAG 2.2 violations (list each with severity: Critical/Major/Minor)\n2. Color contrast failures with specific element selectors\n3. Missing ARIA labels and roles\n4. Keyboard navigation blockers\n5. Screen reader compatibility issues\n\nFormat your response as a structured audit report with an Executive Summary first, then detailed findings grouped by severity. End with a prioritized remediation checklist.",
  "icon": "shield-check",
  "trigger": "on-demand",
  "privacy": "on-device — no data leaves the browser",
  "manifest_snippet": "{\n  \"@context\": \"https://ai.google.dev/gemini-api/docs/chrome-extensions\",\n  \"name\": \"Gemini Nano Accessibility Auditor\",\n  \"version\": \"1.0\",\n  \"model\": \"gemini-nano\",\n  \"trigger\": \"on-demand\",\n  \"privacy\": \"on-device\"\n}"
}