WCAG Regression Diff Auditor

Compare two buyer-supplied axe-compatible JSON snapshots and receive one deterministic report of new, resolved, and persistent violation occurrences.

This Actor does not crawl websites, execute a browser, modify a site, provide legal advice, or certify WCAG compliance.

What this Actor does

• Accepts one baseline and one current snapshot.
• Diffs unique rule ID + target fingerprint occurrences.
• Groups new, resolved, and persistent findings by axe rule.
• Returns only allowlisted rule metadata and SHA-256 selector fingerprints.
• Removes raw HTML, failure summaries, screenshots, page URLs, and arbitrary source payloads.
• Enforces 200 rules and 5,000 node occurrences per snapshot before diffing.

Input

Each snapshot must contain a violations array compatible with the relevant axe result fields:

{
  "comparisonLabel": "release-1-to-release-2",
  "baseline": {
    "violations": [
      {
        "id": "button-name",
        "impact": "serious",
        "help": "Buttons must have discernible text",
        "helpUrl": "https://dequeuniversity.com/rules/axe/4.10/button-name",
        "tags": ["wcag2a", "wcag412"],
        "nodes": [{ "target": ["#save-button"] }]
      }
    ]
  },
  "current": { "violations": [] }
}

Do not include secrets. You are responsible for ensuring that submitted snapshots are authorized and appropriately minimized.

Output boundary

The default dataset receives one report. It contains counts, rule IDs, impact/help metadata, tags, and one-way selector fingerprints. It never returns the original selectors, node HTML, page URLs, failure summaries, screenshots, or raw snapshots.

A target fingerprint helps establish whether the same rule-target occurrence persists. It is not a DOM locator and cannot be reversed by this Actor.

Limits

• Exactly two buyer-supplied snapshots per run.
• Maximum 200 violations/rules and 5,000 node occurrences per snapshot.
• Maximum 2 MB serialized input.
• No network requests, browser execution, proxies, APIs, LLMs, or credentials.
• One output item and at most one draft PPE event per run.

Pricing draft

The private draft models $0.20 per delivered regression report. It is not active pricing. The 20% platform share is a sensitivity assumption only; current Apify terms and measured cloud cost must be verified before any PPE activation.

Interpretation

A regression report is evidence about differences between two supplied scanner outputs. It does not prove accessibility, legal compliance, conformance level, severity prioritization, or absence of issues.