Find accessibility problems. Verify that they were fixed.
Scan a deployed webpage for automatically detectable accessibility issues, receive prioritized remediation guidance, and rescan to confirm changes. Shield uses axe-core and clearly separates automated findings, AI-assisted proposals, and checks that still require human judgment.
New to APIs? See where Shield fits and copy the setup prompt →
Automated and AI-assisted testing cannot determine complete WCAG conformance on their own - some criteria require human judgment. Shield distinguishes automatically detected findings, AI-assisted remediation proposals, and items flagged for manual review - it never certifies legal compliance.
- Engine
- axe-core
- Evidence
- Page grounded
- Resolution
- Verified by rescan
One request. Real data back.
Request
curl -X POST https://shield-api.3stoneai.com/api/v1/shield/scan \
-H "Authorization: Bearer sk_shield_..." \
-H "Content-Type: application/json" \
-d '{"url":"https://example.com"}'Response
{
"ok": true,
"url": "https://example.com",
"violations": [
{
"ruleId": "image-alt",
"impact": "critical",
"description": "Images must have alternate text",
"wcagCriteria": ["wcag2a", "wcag111"],
"occurrences": 3
}
],
"report": {
"summary": "...",
"prioritizedActions": [ /* AI-generated, plain-language, prioritized */ ]
},
"usage": { "callsUsedThisPeriod": 12, "callsLimitPerPeriod": 500 }
}Every violation is real and traceable to an actual axe-core rule - never invented. The report object is AI-generated from those real findings (Claude), never claims "full compliance," and costs nothing extra - it's included in every call.
This is the quickstart. Full API reference covers every endpoint - scan history, AI-proposed remediations, rescan-based verification, and regression detection.
Five real steps, not five buzzwords
Scan
Real axe-core scan of a live URL. POST /v1/shield/scan.
Understand
Every finding tagged with WCAG criteria, level, and whether it's automated or needs manual review.
Fix
Request an AI-proposed remediation grounded in the real HTML - explanation, diff, confidence, honest risk flags.
Verify
A real rescan confirms whether a specific finding is actually gone - not a self-reported "fixed" flag.
Prevent
Compare scans over time to catch new violations and regressions before they ship - wire it into CI.
Automated scanning and AI assistance find and help fix a real, meaningful subset of accessibility issues - they do not determine complete WCAG conformance on their own. Some criteria require a human reviewer. Shield is evidence and engineering assistance, not a compliance certificate.
Built for people building on top of it, not end users
Web agencies
Offer real accessibility auditing to your own clients without building or maintaining a scanner.
Website builders & QA tools
Add real WCAG checking to your own product's pipeline with one integration.
Compliance & legal-tech platforms
Ground your own compliance features in real scan data instead of a static checklist.
Flat monthly plans, real call caps
No per-seat pricing, no sales call. Pick a plan, get your key in the confirmation email.
What this is - and isn't
Real automated WCAG 2.1/2.2 scanning via axe-core, an AI layer that proposes grounded code fixes, and real rescan-based verification - not a proprietary black-box score, not a fabricated compliance certificate. Every finding is labeled by how it was actually determined: automated (a definite axe-core violation), or needs review (a check axe-core itself couldn't resolve automatically) - never silently marked as passing just because nothing was flagged. Rate limits and usage are enforced per key - see your dashboard email for your current usage.
Already a customer? Open your dashboard ormanage your subscription. Wondering whether you need an API instead of a free scanner? Here's an honest comparison. Curious how the scanner itself is built? Read the engineering writeup.