Skip to content

v1.0.2 - Evidence-led accessibility repairs

Choose a tag to compare

@deepakkamboj deepakkamboj released this 29 Sep 21:09
· 11 commits to main since this release

Accessibility investigation and verification

  • Strengthen all seven accessibility skills with evidence-led diagnosis, bounded repairs and explicit non-fix outcomes.
  • Trace multi-repository applications through shared controls to the exact loaded dependency version.
  • Distinguish Fluent UI v8 from v9; require same-version reproduction before assigning upstream ownership, and approval for upgrades or migrations.
  • Verify the original scenario on the exact candidate build with required browser/assistive-technology evidence.
  • Correct static-analysis false positives, toggle-button semantics and manual-only AT regression requirements.
  • Separate verified fixes, incomplete work and carried outcomes in reports.

Automation and packaging

  • Add configuration-driven scheduled fixing and independent validation templates.
  • Include automation templates, contract checks and evaluation fixtures in the npm package.
  • Synchronize npm, plugin and marketplace versions to 1.0.2.

Validation

  • Structural validation: 36 skills and 20 agents.
  • All 12 instruction contract checks pass in the repository and extracted npm tarball.
  • 22 behavioral evaluation fixtures are included; they have not been executed against a model or live product/AT environment.

Upgrade

Install with npm install -g developer-assistant-skills@1.0.2.
Refresh the installed plugin and re-import saved Copilot automation prompts separately.
Review the new configuration-driven prompt settings and version-2 handoff before enabling scheduled writes.

Rollback

Pin developer-assistant-skills@1.0.1 and restore the previously reviewed scheduled prompt configuration if needed.
This release does not merge its release branch into main automatically; the version and packaging commit remains available on release/v1.0.2 for human review.