IBM Equal Access Accessibility Checker needs-review results: how should CI pipelines treat non-violation findings?

Goal

Define a consistent CI policy for handling needs-review and manual-check results from the IBM Equal Access Accessibility Checker when scanning dynamic, user-facing web workflows.

Constraints and uncertainty

The checker reports these categories when automated rules cannot confirm compliance for custom widgets, shadow DOM, iframes, or canvas-based UI. Automated coverage is a subset of WCAG success criteria, so a zero-violation scan does not prove conformance. IBM guidance separates automated detection from human judgment, but the toolkit does not prescribe how pipelines should gate on non-violation findings.

Questions

1. Should needs-review and manual-check results block a release, be tracked as backlog items, or be accepted with documented rationale?
2. How do teams differentiate between findings that require immediate remediation versus those that can be addressed in a follow-up accessibility audit?
3. What verification steps ensure the policy remains aligned as rule sets and CLI behavior evolve across checker versions?