equall-cli

v0.2.1

Published

10 hours ago

Open-source accessibility scoring CLI — aggregates axe-core, eslint-plugin-jsx-a11y, and more into a unified score.

0High
0Medium
0Low

gotabird

accessibility a11y wcag scoring cli axe-core eslint audit

Equall

Open-source accessibility scoring for dev teams.

Equall aggregates axe-core, eslint-plugin-jsx-a11y, text-readability, and more into a single WCAG score — and delivers an honest, per-criterion conformance verdict.

One command. Real score. No config.

npx equall-cli scan .

  ◆ EQUALL — Accessibility Score

  Summary
  1 file scanned  ·  3 WCAG violations  ·  0 best-practice recommendations
  Coverage  Level A   17/31 checked (55%)  ·  3 failing
            Level AA  25/55 checked (45%)  ·  3 failing

  …grouped WCAG violations, each with how-to-fix and its scanner source…

    54.63    WCAG 2.2 · score is a trend indicator
  3 A/AA failures among the 25 criteria automatically verified (30 not evaluated).

  WCAG 2.2 Support Summary — AA target · automated basis only
  ✓ Supports (automated) 20   ✕ Does not support 3   ○ Not evaluated 32
  What each verdict means → equallscan.com/docs/verdicts

What is Equall?

Accessibility tools today usually 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 sits in that gap. It wraps existing open-source scanners and adds what they're missing: a per-criterion conformance verdict you can actually act on, and a score to track your trend over time.

Aggregator, not a reinventor — wraps axe-core, eslint-plugin-jsx-a11y, readability, and more. We don't rewrite the rules; we unify and de-duplicate their results.
Framework-aware — scans HTML, JSX/TSX, Vue, Svelte, and Astro. .astro is scanned full multi-engine (axe-core + jsx-a11y via astro-eslint-parser + readability), not axe alone.
Per-criterion verdicts — for every WCAG success criterion of your target level: Supports (automated), Does not support, or Not evaluated. It's the backbone of the report; the verdict reference says exactly what each claims — and what it doesn't.
Score is a trend, not a grade — a 0–100 number to watch move over time. It motivates; it certifies nothing. No fake "100% Meets WCAG" badges here.
Speaks the legal standard — --standard wcag21 renders the WCAG 2.1 AA view cited by the EU Web Accessibility Directive / EN 301 549; wcag22 is the default. Same scan, same score — only the criteria set changes.
Honest about coverage — automation covers a subset; the rest is Not evaluated (needs a rendered check or manual review). Page-level rules (landmarks, skip link, <html lang>) are reported as "not verifiable on this scan" when you scan components or partials — not as false violations.

Install

npm install -g equall-cli   # or run on demand with: npx equall-cli scan .

Usage

equall scan .                      # score the current directory (WCAG 2.2, Level AA)
equall scan . --standard wcag21    # WCAG 2.1 AA — the public-sector legal bar
equall scan . --level A            # target a different conformance level
equall scan . --verbose            # full per-criterion support table
equall scan . --json               # machine-readable output for CI / tooling
equall scan . --min-score 90       # CI gate: exit 1 if the score is below 90
equall --help                      # all commands and options

A successful scan always exits 0. Pass --min-score <n> to fail a pipeline when the score drops below a threshold. Criteria above your target level (e.g. AAA under the default AA target) are advisory and never count against the score.

Programmatic use

import { runScan } from 'equall-cli'

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

console.log('Score:', result.score)
console.log('Verdicts:', result.criterion_conformance)

ScanResult carries criterion_conformance (the per-criterion verdicts), coverage (what was actually exercised), summary, standard, and engine_version / score_model version stamps so results stay comparable across releases.

Disk scans also carry routes — the URL patterns the project's file-based routing defines (Next.js App/Pages Router, Astro, plain .html), each as { pattern, file, framework, dynamic } with dynamic segments keeping their bracket syntax (/products/[slug]). The field is tri-state: absent when detection was not attempted (in-memory input), [] when the tree had no supported routing — both declared on diagnostics. Routes are inventory metadata only and never affect the score, verdicts, or coverage.

Documentation

Full guide — scanners, the scoring model, the verdict reference, ignoring issues, CI gates and the API: https://equallscan.com/docs

Published

Vulnerabilities

Links

Maintainers

Keywords

Readme