Accessibility scans with Lighthouse and AI
Each scan loads your page in Chromium, runs Google Lighthouse's accessibility audit, then has Claude, Anthropic's AI model, review the screenshot, accessibility tree and HTML. In 1 to 5 minutes you get a report of WCAG issues with suggested fixes.
Core scanning
What runs on every scan, on every plan.
AI review with Claude
Claude, Anthropic's AI model, reviews the screenshot, accessibility tree, HTML and Lighthouse results. It can flag problems that take judgment, like alt text that doesn't describe the image or a label that doesn't say what to enter.
- Checks whether text alternatives and labels make sense
- Reads the screenshot for visual problems, like hard-to-read text
- Suggests a code fix for each issue
Playwright automation
Playwright loads the page in Chromium, so content rendered by JavaScript is included. It captures a full-page screenshot, the accessibility tree and the HTML.
Lighthouse integration
Google Lighthouse runs its rule-based accessibility audit on every scan. Claude gets its results as input, and the report shows the Lighthouse score.
Visual issue highlighting
The full-page screenshot is marked up with boxes around the issues, colored by severity. Switch between the original and annotated views, and use the magnifier to look closer.
Results and reporting
What's in the report and what you can do with it.
WCAG criteria on every issue
Each issue names the WCAG success criterion it fails and that criterion's level (A, AA or AAA).
Issue prioritization
Each issue has a severity (Critical, High, Medium or Low) and a matching fix priority, from Immediate to Low Priority, so you can start with the most serious problems.
Print or save as PDF
Print the report, or save it as a PDF from your browser's print dialog. You can choose whether to include the header and screenshot.
Shareable reports
Pro and Team
Create a public link to a report so a client can read it without an account.
Workflow and automation
Keep scanning the same pages over time.
Scheduled scans
Hobby and above
Scan a URL daily, weekly or monthly. Scheduled scans run at 6 AM UTC, and you can get an email with the score when each one finishes.
- Daily
- Every day
- Weekly
- On the day of the week you pick
- Monthly
- On a day you pick, from the 1st to the 28th
- Runs without you starting it
- Shows whether the score changed after a deploy
- Up to 10 schedules per project and 50 in total
- Each run uses one included scan
Project organization
Group scans into projects, such as one per site or one per section of an app.
Scan history
Every scan stays in its project with its date and score, so you can see whether the score went up after a round of fixes.
Email notifications
Hobby and above
Get an email with the score when a scheduled scan finishes.
API and CI/CD integration
Start scans from scripts and CI pipelines on Hobby, Pro and Team.
Developer REST API
Hobby and above
Start a scan with a POST request, then fetch the results as JSON: issues, scores, statistics and screenshot links. The full API documentation is available after sign-up.
- Works from GitHub Actions, GitLab CI or any script that can make HTTP requests
- Optional webhook when a scan completes or fails
- Scan public preview deployments before they're merged
Webhook notifications
Pass a
webhookUrlwith a scan and AccessHawk POSTs a signed JSON payload to it when the scan completes or fails.CI/CD pipeline ready
Add a pipeline step that calls the API and reads the results. Your script decides when to fail the build, for example on any Critical issue or a score below 80.
Start with 2 free scans
Create an account to scan your own pages. Each scan takes 1 to 5 minutes. No credit card required.