[sbc1-code]

When unavailable waits were hidden, searching for an existing crossing could show “No crossings match” and remove the control needed to show it again. Keep the results toolbar available when every matching card is hidden, distinguish hidden matches from a genuine no-match search, and expose the disclosure state and controlled results element. Hiding the final card retains keyboard focus on the recovery action.

Three files only: Dashboard and a focused browser regression wired into the existing route suite. Ref #62; the broader mobile-selection issue remains open.

Validation:

• Local production build and full npm test PASS on source tree 97ab3b3b348b5b0378311eaf1a892bd2de606cc5, unchanged in this head: 133 unit tests and 418 manifest browser navigations (179 canonical + 29 aliases + analytics, each desktop/mobile).
• Dedicated EN/ES QA at 360×800, 390×844 and 1440×900: hidden-only search/region recovery, keyboard Enter/Space and focus, crossing detail/Back, absent waits, genuine zero, closure, failed refresh/recovery, stale input and genuine no-match search. Before/after captures use the same unchanged recorded snapshot.
• Entry JS 61.81/62 KiB gzip; CSS 19.09/20. Budgets unchanged.
• Two pre-existing moderate CSS build-chain package entries remain; zero high/critical. Package, lockfile and audit policy unchanged.

No feed, wait values, hours, freshness thresholds, ads, tracking or route generation changes. Rollback is a revert of this single three-file commit. Required CI and final scope review precede release.

[sbc1-code]

Published and functionally verified on https://borderpulse.com.

Merge/source: d214944c11d9b7da0935e4dde2d47b02f2c02e64, reviewed head cb81ad275652b0e0b13cba5c599f97e84d66785b. The actual merge tree exactly matches the freshly tested integration with current collector data; the three-file scope is unchanged.

Automatic production workflow 39996248107 SUCCESS. Vercel dpl_2NdURJ4KpZUQ6GDcP8spHcoYjrM5 READY at the exact deployment, with apex/www aliases confirmed. The public publication manifest served the exact merge SHA before and after browser QA beginning 2026-10-23T17:40:22Z; loaded entry-module bytes and SHA256 were retained.

• Build and full release checks on the deployed artifact: 133 unit tests and 406 manifest navigations PASS, including both EN/ES recovery regressions. Entry JS 61.95/62 KiB gzip; CSS 19.11/20. Budgets unchanged.
• Native EN/ES QA at 1440×900, 360×800 and 390×844: hide unavailable waits, search the existing Progreso report, recover using Tab/Enter and Space, retain focus when hiding the final card, open its detail page and return with Back. Actual API response was HTTP200 with an absent standard wait; it remained unavailable rather than becoming zero.
• Separate controlled reports against the served frontend passed missing/zero/closed, failed refresh, recovery, stale input, region filtering and genuine no-match cases. These are test fixtures, not claims that the live feed failed. Twelve native captures; no captured page errors; own headless browser closed and shared browser unused.

No dependency, ads, tracking, hours, wait-value, freshness-rule, credential or security-setting changes. Two pre-existing moderate dependency entries remain; this is bounded testing, not a complete accessibility/security certification. Broader issue #62 remains open.

Rollback is a reviewed git revert -m 1 d214944c11d9b7da0935e4dde2d47b02f2c02e64 from then-current main through the normal checks/publication workflow, preserving later collector work. Inverse patch checked; not applied.