3 Agents. 14 Days. 0 → 95% Accessibility Coverage | Playwright + AI for Automated WCAG Auditing

axe-core catches only 34% of WCAG violations. The other 66% require contextual
judgment — focus traps, error message associations, reading order mismatches.
We gave that judgment to an AI agent. Here's exactly what we built, what it
found, and why it had been shipping for 8 months undetected.

──────────────────────────────────────────
🔍 WHAT YOU'LL LEARN IN THIS VIDEO
──────────────────────────────────────────
✅ Why axe-core alone leaves 60–70% of WCAG violations uncaught
✅ 3-agent pipeline design: Crawler → Auditor → Tracker
✅ How Claude Sonnet evaluates focus traps, aria-describedby, and touch targets
✅ Deduplicating 847 violations into 94 actionable Jira tickets
✅ Honest caveats: 12% false positive rate & what the pipeline still can't catch

──────────────────────────────────────────
⏱️ TIMESTAMPS
──────────────────────────────────────────
00:00 – The accessibility testing gap nobody solves
00:28 – Why axe-core catches only 34% of WCAG violations
00:55 – Pipeline overview: 3 agents, 1 data flow
01:20 – Agent 1: The Crawler — route discovery & evidence collection
01:50 – Agent 2: The Auditor — 6 WCAG categories LLMs evaluate better than rules
02:30 – Agent 3: The Tracker — deduplication & Jira ticket prioritisation
03:00 – What we found that had been shipping for 8 months
03:25 – Results: 847 violations, 94 tickets, 23 Level A blockers
03:45 – Honest caveats & what this pipeline still misses
04:05 – Call to action / subscribe

──────────────────────────────────────────
📊 RESULTS AT A GLANCE
──────────────────────────────────────────
- Automated accessibility coverage: 0% → 95% of routes
- WCAG violations surfaced in first scan: 847 across 94 routes
- Unique Jira tickets: 94 (deduplicated from 847 instances)
- Critical Level A violations found: 23 (all assigned within 2 hrs)
- axe-core catch rate vs full audit: 34% → 89%
- Regression detection: quarterly manual → every CI push

──────────────────────────────────────────
🛠️ TECH STACK USED
──────────────────────────────────────────
- n8n (pipeline orchestration)
- Playwright + @axe-core/playwright (browser automation + rule checking)
- Claude Sonnet (contextual WCAG reasoning)
- Jira (ticket creation & lifecycle management)

──────────────────────────────────────────
📌 WHO THIS IS FOR
──────────────────────────────────────────
Senior SDETs and QA Engineers who want to go beyond axe-core, build AI-powered
accessibility pipelines, and create compliance coverage that actually holds up
in production — without hiring a manual auditor.

──────────────────────────────────────────
🔗 CONNECT WITH ME
──────────────────────────────────────────
📝 Medium (deep-dive articles): https://medium.com/@krohit0389
💼 LinkedIn: https://www.linkedin.com/in/rohit-kshirsagar-sdet/
▶️ YouTube: https://www.youtube.com/@RohitKshirsagar-c1y

──────────────────────────────────────────
🔔 If you're an SDET building AI-powered QA pipelines, subscribe — I post
real implementations with real numbers, every week.
──────────────────────────────────────────

#AITesting #Accessibility #WCAG #Playwright #AIAgents #SDET #QAAutomation
#axecore #n8n #ClaudeAI #TestAutomation #AccessibilityTesting

Видео 3 Agents. 14 Days. 0 → 95% Accessibility Coverage | Playwright + AI for Automated WCAG Auditing канала Automate & Elevate