Machine stage
- Required PDF structures and properties
- Programmatically inspectable tag relationships
- Validator profile and engine versions
- Deterministic failures with object evidence
PDF/UA-1 validation API
Run machine-verifiable PDF/UA-1 checks before a PDF leaves the publishing workflow. The API records the validator and profile version, normalizes structural failures, and keeps reading order and meaning in a separate human-review queue.
For PDF production lines, archives, and high-volume document publishers
API profile
pdfua-1-machine@1.2.0Illustrative request
202 AcceptedPOST /v1/scans
Idempotency-Key: board-pack:2026-08:v12
{
"asset": {
"type": "upload",
"upload_id": "upl_board_pack_v12"
},
"profiles": [
"pdfua-1-machine@1.2.0"
],
"metadata": {
"document_id": "board-pack:2026-08:v12"
}
}queued
id: scn_01J…profile resolved- Target
- ISO 14289-1:2014 machine checks
- Validator
- Version retained in result
- Limit
- Machine validation is not complete accessibility
Turn validator output into a repairable document finding.
AccessPreflight keeps the native rule lineage while adding a stable fingerprint, document scope, evidence locator, and source-oriented next step.
- Rule
- Native validator rule + normalized rule ID
- Scope
- Document, page, object, or relationship
- Observed
- Missing or invalid structural condition
- Profile
- pdfua-1-machine@1.2.0
- Engine
- Pinned veraPDF release identifier
- Follow-up
- Manual reading order and meaning checks
Keep PDF/UA-1 preflight inside the document publishing loop.
The asset retains one external identity from upload through repair and re-scan, so editors do not reconcile a separate checker workspace.
- Upload
Reference the source asset
Complete a validated PDF upload.
- Preflight
Resolve PDF/UA-1
Capture machine-verifiable conditions.
- Repair
Return object evidence
Route findings to the authoring source.
- Rescan
Compare the same asset
Keep repaired and open work visible.
The publishing decision follows the asset, not the scanning UI.
Keep machine validation and document review as two named stages.
A clean machine result is useful evidence, but it cannot judge whether the tagged structure communicates the intended document.
Human stage
- Logical reading order
- Alternative-text meaning and context
- Heading and table semantics in use
- Assistive-technology reading experience
Inspect the policy behind the API profile.
Before PDF/UA-1 becomes a production gate, review its source mapping, supported input, automation boundary, and required human checks.
Put the PDF/UA-1 API in the path you already own.
Start with one asset flow, pin the profile, and keep evidence and unresolved human work attached to every result.
Check a website freeCreate an account
Automated preflight is not an accessibility certification or legal determination.