Equall

Open-source accessibility scoring for dev teams. Aggregates axe-core, eslint-plugin-jsx-a11y, text-readability, and more into a unified score.

One command. Real score. No config.

equall scan .

  ◆ EQUALL — Accessibility Score

  56   ~A   WCAG 2.2

  POUR Breakdown
  P Perceivable    ██████████████████░░ 89
  O Operable       ███████████████░░░░░ 76
  U Understandable ░░░░░░░░░░░░░░░░░░░░ n/a
  R Robust         █████████████████░░░ 85

  Summary
  33 files scanned  ·  15 WCAG violations · 19 best-practice issues
  2 critical  11 serious  19 moderate  0 minor
  2 issues ignored via equall-ignore
  Score 56/100  ·  18/32 Level A criteria checked (56%)  ·  5 failed
                   29/56 Level AA criteria checked (52%)  ·  6 failed

  ⓘ You're failing 5 Level A criteria (1.3.1, 2.1.1, 2.4.2, 2.4.4, 4.1.2).
    Level A is the legal minimum — fix these first.

  Best Practices
  ● region — 26 issues  Landmarks help screen reader users navigate page sections
    src/app/layout.tsx
    src/components/Navigation.tsx
    ... and 24 more (use --verbose to see all)

  Scanners: [email protected] (23 issues), [email protected] (13 issues)
  Completed in 0.8s

Install

npm install -g equall-cli

Or run directly with npx:

npx equall scan .

Why Equall?

Accessibility tools today fall into two camps: dev tools that show violations without context (axe, Lighthouse), and enterprise platforms that cost $75K+/year (Deque, Siteimprove). Nothing in between.

Equall aggregates existing open-source scanners and adds what's missing: a score, a trend, and a POUR breakdown that tells you where to focus.

• Aggregator, not reinventor — wraps axe-core, eslint-plugin-jsx-a11y, and more. We don't rewrite rules, we unify results.
• Score 0-100 — weighted by severity, grouped by WCAG criterion, with a conformance level (A / AA / AAA).
• POUR breakdown — see which accessibility principle (Perceivable, Operable, Understandable, Robust) is your weakest.
• Zero config — point it at a folder, get a score. No setup, no account, no signup.
• Honest about coverage — we tell you exactly which criteria we test and which need manual review. No "100% compliant" bullshit.

Usage

Scan a project

equall scan .                    # current directory
equall scan ./my-app             # specific path
equall scan . --level A          # target Level A only
equall scan . --level AAA        # include Level AAA criteria

JSON output

equall scan . --json             # pipe to other tools
equall scan . --json > report.json

Options

equall scan . --include "src/**/*.tsx"
equall scan . --exclude "**/*.stories.*"
equall scan . --verbose              # show all files per best-practice rule
equall scan . --show-ignored         # show ignored issues
equall scan . --show-manual          # list untested criteria needing manual review
equall scan . --no-color             # disable colored output

What gets scanned

Equall discovers and scans: .html, .htm, .jsx, .tsx, .vue, .svelte, .astro

It automatically skips: node_modules, dist, build, .next, test files, stories, and respects .gitignore.

Scanners

| Scanner | What it checks | WCAG criteria covered | |---------|---------------|----------------------| | axe-core | HTML structure, ARIA, landmarks, forms, media | 24 | | eslint-plugin-jsx-a11y | JSX/React-specific a11y patterns | 16 | | readability | Required reading level of text content (Flesch, ARI, SMOG, etc.) | 1 (WCAG 3.1.5 AAA) |

Readability scoring

The readability scanner checks WCAG 3.1.5 (Reading Level) — text should be understandable at a lower secondary education level (approximately Grade 9).

It runs 6 formulas (Flesch-Kincaid, Coleman-Liau, ARI, Gunning Fog, SMOG, Dale-Chall) and uses the median grade across all formulas to reduce single-formula bias. If the median exceeds Grade 9, an issue is reported with the full breakdown.

• Scans .html and .vue files only — JSX/TSX excluded because regex extraction captures className attributes and {expressions} that pollute scores
• Skips files with fewer than 30 words (formulas are statistically invalid on short text)
• Skips non-English files (detected via lang attribute) — formulas are English-calibrated

Scoring

The score (0-100) is designed to be fair and scalable, whether you are scanning 5 files or 5,000. It is calculated using a density-based asymptotic algorithm:

• Severity Weights: Violations add a penalty based on severity (critical: 10, serious: 5, moderate: 2, minor: 1).
• Criterion Caps: Penalties are capped at 15 points per WCAG criterion to prevent a single recurring issue from entirely destroying the score.
• Density Scaling: Total penalties are scaled down logarithmically based on the number of files scanned (1 / (1 + log10(files))). A large repository is allowed more absolute errors than a tiny project for the same score.
• Exponential Curve: The final score follows an asymptotic decay curve (100 * exp(-k * penalty)). It drops quickly for the first few errors but smoothly approaches 0 without ever hitting it brutally, leaving room to measure progression.

POUR Breakdown

The POUR metrics (Perceivable, Operable, Understandable, Robust) strictly follow the same scoring logic, isolated by principle, providing an independent 0-100 score for each accessibility pillar.

Conformance Level

Conformance (A / AA / AAA) is evaluated strictly against your --level target. If you target AA, any AAA rules incidentally flagged by the scanners will not downgrade your conformance status.

Ignoring issues

Some issues are false positives (e.g. an orphan <li> in a component that's always rendered inside a <ul>). Suppress them with inline comments:

// equall-ignore-next-line
<li>{item.name}</li>

// equall-ignore-next-line jsx-a11y/alt-text
<img src={logo} />

{/* equall-ignore-next-line */}
<div onClick={handler}>...</div>

<!-- equall-ignore-next-line -->
<img src="decorative.png" />

Or use the CLI to inject/manage comments without opening the file:

equall ignore src/Modal.tsx                             # ignore an entire file
equall ignore src/Modal.tsx:89                          # ignore all rules at line 89
equall ignore src/Modal.tsx:89 jsx-a11y/alt-text        # ignore a specific rule
equall ignore .                                         # list all ignores
equall ignore --remove src/Modal.tsx:89                 # remove an ignore
equall ignore --remove src/Modal.tsx                    # remove all ignores in a file
equall ignore --clear                                   # remove all ignores

Ignored issues are excluded from the score. Use equall scan . -i to show them, or --json to get them with "ignored": true.

Programmatic API

import { runScan } from 'equall-cli'

const result = await runScan({
  path: './my-project',
  level: 'AA',
})

console.log(result.score)              // 73
console.log(result.conformance_level)  // 'A'
console.log(result.pour_scores)        // { perceivable: 90, operable: 65, ... }
console.log(result.issues.length)      // 12

Exit codes

• 0 — score >= 50
• 1 — score < 50 (useful for CI gates)
• 2 — scan error

Contributing

Issues and PRs welcome — https://github.com/GotaBird/equall/issues

License

MIT