NCSC Switzerland, Im Fokus

https://www.bacs.admin.ch/de/im-fokus

Swiss NCSC focus topics (German). Translate findings to English in the brief. | 2026-05-08 audit: WebFetch returned 5 dated posts with 2026-05-07 latest. | 2026-06-20 full audit (v2.62): live=Y, drill=Y. FETCH → webfetch (listing) https://www.ncsc.admin.ch/ncsc/de/home/aktuell/im-fokus.html then webfetch the per-post .html for body (both work). AVOID: German-language, translate to English in brief. No bridge needed (admin.ch resolves under the routine UA).. | 2026-07-05 admiralty audit: A (HIGH->A) status active->active, Swiss national cyber authority first-party content (home authority); live via WebFetch, latest post 30.06.2026. Justified A: primary government authority for its own jurisdiction. | 2026-08-17: health sweep flagged needs-demote (unreachable). Contention artifact, not a recipe defect, a serial re-probe through the generic bridge returned the full 105 KB 'Im Fokus' listing with dated entries, and a research pass read the same page by plain WebFetch earlier in the run. NOT demoted; fetch_method stays webfetch. | 2026-08-23: NCSC.ch renamed to BACS (Bundesamt fuer Cybersicherheit) and relaunched on bacs.admin.ch around 2026-08-20. The legacy ncsc.admin.ch/ncsc/de/home/aktuell/im-fokus.html path 301-redirects and serves the identical page (verified byte-identical this run), so this is a canonical-URL correction, not a dead source. FETCH -> the page is a Nuxt SPA: WebFetch renders it; the bridge's direct GET returns the JS shell only. | 2026-09-13 quality audit (G2 spot-check, carry-forward watch item 2 from the 2026-09-06 report): RECIPE WAS BROKEN AND IS NOW FIXED. Both BACS pages are Nuxt SPAs that return an EMPTY shell to the recorded transport; the failure is silent, which is why the record stayed green at 7 consecutive quiet periods while contributing nothing since 2026-06-18. Reachability was never the problem; readability was. VERIFIED WORKING this run: `python3 tools/fetch_source.py jina https://www.bacs.admin.ch/de/im-fokus` hydrates the page and returns the dated post list. fetch_method webfetch -> jina on that basis. Cost tradeoff, stated so a later fire can re-judge it: jina spends metered reader credit and a dead pool is a normal condition, so when the pool is empty this page is unreadable and that is a disclosed gap, not a demotion, the Cyber Security Hub (ncsc-ch-security-hub, a free API) remains the operational/sector surface and is unaffected. In-window content check: one post (08.09.2026, 'Woche 36', analog-skimming social-engineering awareness), reachable once hydrated, but it does not clear this deployment's technical-depth bar, so the quiet periods were only PARTLY a transport defect. | 2026-09-27 audit: fetch_method moved jina -> bridge. The jina pin dated from a fire that concluded only the reader hydrates this page; that is no longer true and the reader pool has been empty for two windows (0 live keys), so the page was unreadable by its pinned transport while a working direct one existed. Verified this fire: `python3 tools/fetch_source.py extract <url>` returns the full dated item list via trafilatura-direct, no metered credit. Only fetch_method and these notes changed: the audit probed the transport, it did not fetch-and-use the page for an entry, so last_successful_fetch and the quiet counter are left where the fires put them. | 2026-10-05 S2: `url https://www.bacs.admin.ch/de/im-fokus` returns raw HTML whose cards are `<span class="meta-info__item">D. Monat YYYY</span> ... <h3 class="">Title</h3>` (76 dated focus items); the `extract` view stays undated. | 2026-10-06 (S2): the BACS weekly review is published on Tuesdays (weeks 37 to 39 appeared on 15, 22 and 29 September); 26w40 was not yet out at 04:22Z on 2026-10-06. | 2026-10-13 (S2): the working card parse splits on class="card__content" and reads meta-info__item per card. | 2026-10-22 (S2): recipe: the first `url https://www.bacs.admin.ch/de/im-fokus` call falls through to the metered jina reader (no hrefs); `url --direct` returns the raw HTML with the card hrefs and `extract <card href>` reads each body. | 2026-10-27 (S2): the card markup is now `class="card card--default card--clickable color--text-900"`, so a split on the string with its closing quote returns zero cards; split on `class="card card--default card--clickable` without the closing quote, title in `class="card__title"`, date in `meta-info__item">`, href is the first `href=`.