a11y checker
Changelog

WCAG keeps moving. AI keeps moving.
So we keep updating.

Support for new AI models, verdict-accuracy improvements, and new features, published here in chronological order. In a field where both the accessibility standard and the judging models keep evolving, this page is where you can confirm the tool is alive and current.

2026-09-18

Measurement layer under the AI verdicts, and a "needs review" status

  • NewFor text contrast (1.4.3), control boundaries (1.4.11), reflow (1.4.10), 200% zoom (1.4.4) and text spacing (1.4.12) the audit container now measures the live page — every visible text element's contrast against its rendered background (including text over images, sampled from pixels), text-field borders, horizontal overflow, clipped text and overlapping blocks at 320px / 640px / with the spacing overrides. A measured violation fails the check with the measured element as evidence even when the AI passed it.
  • NewWhen the AI fails one of these checks but the element it named measures within the threshold, the check becomes Needs review instead of a fail. It is not a pass, it does not count as fixed, and it blocks "conforms" exactly like "unverified" — a person decides. Every result keeps the raw AI verdict, the measurement and the rule that combined them (provenance in the JSON export).
  • ImprovedBorderline pixel measurements (within ±0.25 of the threshold), inactive controls (disabled / aria-disabled) and line-clamped summaries are never turned into a confirmed pass; a clamp that hides materially more text under zoom or spacing overrides is flagged for review.
2026-09-17

Report download (PDF, HTML, Excel) now on sale

  • NewFor ¥980 (tax incl.) per audit you can download the stored result as PDF, HTML and Excel. It is a separate product from credits and never re-runs the AI audit; every member of the workspace can re-download at any time. Viewing the web report remains free.
  • ImprovedDownload access is granted only by the signature-verified Stripe webhook after checking amount, currency, audit and workspace. If generation fails, your purchase stays valid and you can retry.
Mid-September 2026

AI included: no more API keys, credit packs on sale

  • NewAudits now run on our own AI capacity (vision-capable models on Ollama Cloud). The bring-your-own-key (BYOK) model and the provider/model pickers are gone — nothing to paste, no separate LLM bill.
  • NewCredit packs (10 / 50 / 200 credits) can be bought in the dashboard with card payment via Stripe. Paid credits never expire; the monthly free credits stay. See pricing.
  • ImprovedHonest verdicts: a check the AI could not judge is now marked unverified, excluded from the tally and blocks a "conforms" result — it is never treated as a pass any more.
  • ImprovedCredits are reserved, spent and refunded through exactly-once state transitions; purchases are granted only from the signed payment webhook. The dashboard shows the cost of a run before you start it.
2026-07-22

Measured-verdict layer and accuracy benchmark

  • NewAdded a measured-verdict layer — target size (WCAG 2.5.8 / 2.5.5), reflow (1.4.10), text spacing (1.4.12), and focus visibility (2.4.7) are now decided from actual pixel and box measurements before the AI is ever asked. The measured values are kept in the evidence for each verdict.
  • NewIntroduced an accuracy benchmark — detection and false-positive rates are continuously measured against test pages with known, seeded defects. The deterministic and measured layers must hold 100% detection with zero false positives as a release condition.
  • NewJIS X 8341-3:2016 test-result summary — audit results can be summarized (in Japanese) with JIS-covered criteria (the 60 from WCAG 2.0) separated from reference criteria (the 26 added in WCAG 2.1/2.2).
  • NewStronger CI integration — SARIF 2.1.0 output (GitHub Code Scanning compatible) and a regression gate that diffs against the previous run to block only new violations, both added to the CLI.
  • ModelsAdded support for Gemini 3.6 Flash and Gemini 3.5 Flash Lite (the day after Google's announcement). The existing 3.5 Flash and 3.1-series models remain available.
Mid-July 2026

Contact channel and free onboarding

  • NewAdded a contact form (/contact). Submissions are tracked internally and a confirmation email is sent automatically.
  • NewPublished a guide to getting an API key for free (retired in September 2026 with the move to included AI) — covering the Gemini free tier and OpenAI data sharing.
  • ImprovedImproved the accessibility of the landing page itself, reaching a Lighthouse accessibility score of 100 (dogfooding).
Early July 2026

Launch on the official domain, isolated-container audits

  • NewOfficially launched at a11y.effloow.com. Audits now run in an isolated container per URL and keep running even if you close your browser. BYOK API keys live only in container memory and are never stored.
  • NewStarted selling reports (detailed Excel report, VPAT®/JIS report) with credit-card checkout.
  • ImprovedIntroduced a fail-safe design that never consumes a credit for a failed audit (hold and automatic refund).
Have a request, or curious what's next? Send it through the contact form. Updates are published on this page as they ship.