Quality · Accessibility
A11yForge
Ship it accessible. Prove it with a VPAT.
A11yForge is a local WCAG Level A & AA testing workspace for engineers. Set a site up once, then assess it — and re-assess it over time — working the criteria grouped by how you actually test: automated, keyboard, screen reader, visual. Attach evidence to every finding and export a VPAT/ACR. It runs in your browser and keeps everything in local storage — no account, no backend.
A11yForge speeds up the audit — it doesn’t replace it. Automated scanning only covers a slice of the criteria; a real VPAT still needs manual testing and human judgement. It’s a local, single-user workspace focused on the audit itself — not a compliance guarantee or a multi-user platform.
Overview
What is A11yForge?
A11yForge is a quality-engineering tool for accessibility conformance work. It separates the durable thing you test — a product, set up once with its team and default target — from an assessment, one point-in-time evaluation against a chosen WCAG target. A product builds up a history of assessments, so re-testing a release is a copy-and-regrade, not a blank page. Scope is deliberately websites only.
The 56 success criteria — WCAG 2.1 Level A & AA, plus the six added in 2.2 — are organised by testing method rather than spec order, each graded Supports, Partially Supports or Does Not Support with remarks and evidence. axe-core results pre-grade the criteria a scanner can check, and by design the auto-grader never claims a criterion is supported unless it actually exercised it. The result is a VPAT/ACR that explains its conclusions — not a black-box score.
What it does
From portfolio to signed VPAT
Portfolio & re-assessment
Manage a portfolio of sites grouped by organisation. Each product is set up once — team, owner, default conformance target — then carries a full history of point-in-time assessments. Re-assess a release and the header copies over; the issued VPAT stays as the record.
Testing grouped by method
Work the 56 WCAG Level A & AA criteria — the 2.1 set plus the six added in 2.2 — grouped by how you actually test: automated, keyboard, assistive tech, visual, media, pointer. Grade each Supports, Partially Supports or Does Not Support against a target you choose.
axe-core pre-grading
Pull in axe-core results to pre-grade the automatable criteria. Real violations attach as evidence — and, by design, the auto-grader never marks a criterion “Supports” for something it didn’t actually exercise. The scanner handles the obvious; you make the judgement calls.
VPAT / ACR — four editions
Export a simplified internal report, the official ITI VPAT® 2.5, a Section 508 edition, or EN 301 549 (EU) — as printable HTML (Save as PDF) or an editable Word document. Unfinished reports are watermarked DRAFT so they can’t be mistaken for final.
Sign-off & attestation
Sign an assessment to turn it into a dated, frozen attestation — its grades and evidence lock, so the signature can never drift from what it covered. Un-sign deliberately to edit, or copy any finding straight to a Jira ticket for remediation.
Journey Matrix
Chain products into a customer journey and read the combined conformance. Each criterion rolls up to the weakest step — one failing step fails the journey — so the matrix can never over-claim what its parts actually support.
“A VPAT shouldn’t mean a weekend lost in a spreadsheet. Set the product up once, test by method, sign it off — and keep every re-assessment as history, on your own machine.”
Run an assessment
Open A11yForge, set up a product, and take an assessment from first criterion to signed VPAT. Free to use, nothing to install — it seeds a few example products so there’s something to explore straight away.
Launch A11yForge