Verdict reference
What each support verdict means — Supports (automated), Does not support, Not evaluated — and how it maps to the VPAT/ACR vocabulary. The honest anchor for reading a scan.
For every WCAG success criterion of the target level, a scan states one honest, scan-scoped verdict. This page is the authoritative reference for what each verdict asserts — and, just as importantly, what it does not.
Equall's output is the evidence layer, not a VPAT. The engine never emits a formal "Supports" on its own; a documented map (below) translates each verdict to the ITI VPAT/ACR vocabulary, which a human applies during attestation.
What you see (terminal Support Summary)
The terminal groups the verdicts into three buckets:
| Bucket | Meaning |
|---|---|
| ✓ Supports (automated) | Automated checks exercised the criterion and found no failure. An automated basis only — not a claim that the criterion is actually met. |
| ✕ Does not support | At least one automated check failed the criterion. Backed by evidence (the failing findings). |
| ○ Not evaluated | Automation did not establish the criterion on this scan. Splits into three reasons under --verbose. |
The five verdicts
These are the machine values on criterion_conformance (the terminal collapses the three not_* verdicts into "Not evaluated").
| Verdict | Asserts | Does not assert | Evidence / reason |
|---|---|---|---|
fail | ≥1 automated check failed this criterion | — | Failing finding fingerprints (evidence) |
pass_automated | Automated checks ran and found no failure | That the criterion is met (needs manual + assistive-tech review) | — |
not_verifiable_on_this_scan | A page-level rule could not be checked on a fragment | Anything about the composed page | Reason: verify on the rendered / built output |
not_tested_assisted | Partially covered by static analysis (e.g. contrast) | A pass or a fail | Reason: needs a rendered / assisted check |
not_tested_manual | No automated coverage on this scan | Anything | Reason: verify manually |
Every Supports (automated) carries the (automated) qualifier for a reason: automation cannot judge semantic quality. A present-but-useless alt="DSC00423" passes automated checks — the verdict says the automated basis is clean, never that the image is accessible. Confirm with manual + assistive-technology testing.
Mapping to the VPAT / ACR vocabulary
When you build an accessibility statement or ACR, each engine verdict maps to the ITI VPAT term below. This mapping is documented here as the reference (it is not exported from the package — consume criterion_conformance from the result) and is applied with human attestation — the engine never asserts a formal "Supports" by itself.
| Engine verdict | VPAT / ACR term |
|---|---|
fail | Does Not Support / Partially Supports |
pass_automated | Supports — automated basis, pending manual confirmation |
not_verifiable_on_this_scan | Not Evaluated (verify on rendered / built output) |
not_tested_assisted | Not Evaluated (assisted / partial) |
not_tested_manual | Not Evaluated (manual) |
There is intentionally no Not Applicable: applicability requires content judgement a scanner cannot honestly make — it is a human attestation.
Coverage & completeness
The verdicts are emitted for every criterion of the target level, so they always sum to that level's criteria total — no criterion is silently missing. See Output format for the criterion_conformance JSON shape, and Scanning for the terminal Support Summary.