ci: run build-output-diff on all PRs (drop paths filter)

[sanrai]

Removes the paths filter so build-output-diff runs on every PR, not just dependency bumps. Every PR now gets a NO_CHANGE / CHANGED verdict (comparing both main.js and app.css). This closes the gap where non-dependency PRs (CI/tooling, refactors) were reported as SKIPPED because no verdict was available. deps-dev auto-pass is unchanged; human PRs still require review. Note: check-pr-title will fail without an mwpw JIRA scope in the title.

[github-actions]

Agent QA review — interactive + visual diff (advisory, non-blocking)

Last updated Aug 14, 2026, 10:54 PM PDT · PR opened · commit f0a08ff · 1 file changed.

0 open · 0 resolved · visual diff 0.00% · verdict PASS

Open findings

• Smoke test passed ✅

What the agent checked

Reviewed PR #579 on the live CaaS collection at business.adobe.com/resources/main.html.

Context: This PR only touches .github/workflows/output-diff-check.yml (a CI workflow config that removes a paths filter so the build-output-diff advisory check runs on every PR). It contains zero application/frontend code changes. Consistent with that, the automated pixel-diff report showed 0.00% pixels changed, and the loaded diff.png showed no magenta/highlighted regions anywhere on the page — confirming there is no visual delta between the PR build and stable build to inspect.

Since there was no rendering change to focus on, I performed a general functional QA pass on the live Resource Center (CaaS) collection to confirm the page is healthy and nothing is incidentally broken:

1. Page load: main.html loads cleanly, header nav, breadcrumb (Home/Resource Center), "Explore all Adobe resources" heading, and card grid all render correctly with no layout issues.
2. Filter panel (LeftFilters): "Refine Your Results" panel with Search, Products, Content type, Industry accordions works as expected. Clicking "Products" expands the checkbox list correctly (Acrobat, Acrobat Sign, Advertising, Analytics, etc.) with a smooth chevron rotation.
3. Applying a filter: Checking "Acrobat" correctly (a) added an "Acrobat ×" removable pill under the search box, (b) updated the result count from 3796 → 182, (c) showed a badge "1" next to Products, and (d) refreshed the card grid to Acrobat-relevant articles ("Unlock sales advantage with AI-powered workflows", "Explore 5 ways to streamline marketing workflows"). All correct, no stale/mismatched cards.
4. Search input and its "Clear Search filter" button are present and rendered correctly in the filter sidebar.
5. Cards render with images, titles, descriptions, and "Read article" CTA buttons all properly aligned — no truncation, overlap, or broken images observed.
6. run_axe on the filter panel: 0 violations, 22 passes, 0 incomplete — no accessibility regressions.
7. get_console_errors: only one benign message ("Permissions policy violation: unload is not allowed in this document"), which is a browser-level warning unrelated to this PR's CI-only change and not a JS crash/error.

No broken, misaligned, truncated, or low-contrast elements were found. No console errors attributable to app code. The PR itself is CI-configuration-only and does not affect the rendered page, which matches the observed 0.00% pixel diff and clean functional walkthrough.

Verdict: PASS — safe, no regression risk to the live page; changes are confined to CI workflow triggering logic.

PR / stable / diff screenshots + console + axe artifacts in the workflow run.

Review history (1 run)

• f0a08ff · Aug 14, 2026, 10:54 PM PDT · PR opened · passing, no regressions — ci: run build-output-diff on all PRs (drop paths filter)

[github-actions]

🧪 Feature QA review — injected feature test (advisory, non-blocking)

Last updated Aug 14, 2026, 10:54 PM PT · commit 0bd2ed7 · pull_request.

Overall: SAFE (no bundle change)

No injectable feature to exercise — and the deterministic build-output-diff proves the shipped bundle is byte-identical to base after normalization, so this change is safe for what ships.

This PR only changes a GitHub Actions workflow's trigger filter (CI/tooling), with no config/card-data-driven UI behavior to exercise in a rendered page.

Screenshot in the workflow run.

Review history (1 run)

• Aug 14, 2026, 10:54 PM PT · SAFE (no bundle change) · 0bd2ed7 · pull_request

[github-actions]

AI Code Review

Last updated Aug 14, 2026 10:55 PM PDT · PR opened · commit f0a08ff · 1 file changed in PR.

0 open · 0 resolved

Open findings

No open findings. ✅

Review history (1 run)

• f0a08ff · Aug 14, 2026 10:55 PM PDT · PR opened · no changes (0 open) — ci: run build-output-diff on all PRs (drop paths filter)

[sanrai]

Folded into #578; closing this separate PR.