Accessibility Regression Checker (ARC)

A toolchain to automate web accessibility regression checks.
View Demo · Report Bug · Request Feature

About the Project

The Accessibility Regression Checker (ARC) is a database-backed toolchain for auditing and tracking web accessibility compliance using axe-core, Node.js, SQLite, and Puppeteer. While automated testing alone cannot resolve every barrier, ARC serves as an early warning system by highlighting new violations that require deeper manual reviews.

Because accessibility is a continuous process, websites should be audited on a regular schedule. Even a fully remediated site can introduce new issues as it evolves. ARC can run baseline scans (typically taken right after accessibility fixes) to capture a site’s current compliance level. Future scans compare against this baseline to detect any regressions. If new violations appear, ARC sends alerts, prompting stakeholders to initaite a thorough review and remediation by an accessibility expert.

Built With

Getting Started

Prerequisites:

• Install npm:

  npm install npm@latest -g

Installation:

• Clone the repository:

  git clone repo-url

• Navigate to the project directory:

  cd accessibilities-checker

• Install dependencies:

  npm install

(back to top)

Usage

Running Locally:

Run scan-website.js with or without the --baseline flag.

• With --baseline: Saves detected barriers to the database as a baseline.
• Without --baseline: Compares new barriers to the baseline and logs results.

Baseline Scan (run after latest accessibility remediations have been completed):

node inspections/scan-website.js https://example.com --baseline

Scan (run regularly to check for any new accessibility barriers):

node inspections/scan-website.js https://example.com

Alerts & Configuration

The webhook-server.js script helps integrate ARC with external platforms, such as CRMs, email services, or Slack.

Inbound (Scan Requests):

External tools can send a POST request to the /webhook endpoint with:

• url: The website URL to scan.
• isBaseline: Whether the scan should be a baseline scan (true) or a regular scan (false).

Example Request

curl -X POST http://localhost:3000/webhook \
     -H "Content-Type: application/json" \
     -H "x-auth-token: <your-secret-token>" \
     -d '{"url": "https://example.com", "isBaseline": true}'

Outbound (Regression Notifications):

By default, ARC sends regression data to the /regression endpoint after each comparison. This payload can be forwarded to any external service that accepts JSON (e.g., Slack, email, CRMs):

• Slack: Use a Slack Incoming Webhook URL to post violation details to a channel.
• Email: Configure a backend script (e.g., Nodemailer) to forward the JSON payload via email.
• CRMs: Platforms like Salesforce, HubSpot, or GoHighLevel can trigger tasks, notifications, or other workflows using the POST data.

(back to top)

How It Works:

axe-core is an accessibility testing engine that runs within the browser to identify accessibility barriers on web pages, following WCAG guidelines and best practices.

WCAG (Web Content Accessibility Guidelines) is a set of internationally recognized standards for making web content more accessible to individuals with disabilities, ensuring equal access and usability for all users.

Puppeteer is used to programmatically launch a browser, navigate to the target website or local HTML file, inject the axe-core library, and execute accessibility analysis in the browser context.

Key Features

• Scanning and Comparison:

  • inspections/scan-website.js: Scans websites or local HTML files for accessibility barriers.
    • Baseline Scans: Save detected barriers as a reference point for future comparisons.
    • Non-Baseline Scans: Compare scan results to the baseline, detecting regressions such as new WCAG violations.
  • inspections/compare-baseline.js: Identifies accessibility regressions by comparing barriers in the latest non-baseline scan with the baseline. Sends regression details to a webhook for alerts.

• Notification System:

  • webhook-server.js:
    • Inbound: Processes scan requests with URL and baseline parameters. Triggers scan-website.js.
    • Outbound: Sends notifications when regressions are detected, including details about new WCAG barriers.

• Storage:

  • db/init.js: Initializes an SQLite database for storing inspection metadata and barrier details.
    • Inspections Table: Tracks metadata such as scan type (baseline/non-baseline), URL, and environment.
    • Barriers Table: Stores details of detected barriers, including WCAG rules, impact, and affected user groups.

(back to top)

Contact

Julian De Los Prados - jad9679@nyu.edu

Project Link: https://github.com/jdelosprados/accessibility-regression-checker

(back to top)