de:wasserversorger-enrich

Journal entry appended and the pipeline finished cleanly. Final report below.

The scheduled 45-company batch (shard A: 23, shard B: 22) completed successfully — all companies confirmed genuine drinking-water suppliers (0 non-suppliers), with enrichment.jsonl growing 1799→1844 and roster-links.jsonl 20061→20205 after merge. Population coverage held steady at 99.99% (24 residual municipalities, 11,023 people, unchanged as expected since this batch enriched already-claimed companies); analysis-locator recovery was ~37/45 and 8 new host traps were logged (notably new durable 403 blocks on braunfels.de and stadtwerke-hammelburg.de, plus a dead-domain migration wzv-denkendorf.de→wzv-ei.com flagged for a manual domain update). Two items worth human attention: a possible Burgau/Stadtwerke Günzburg supply-area overlap on Limbach, and Gemeindewerke Morsbach missing its legally-required analysis publication.

build_enrich_batch_excl.mjs 45 → 0 pending_domain (queue still empty), batch drawn purely by population priority. Split shard A (23) / shard B (22), each dispatched to one top-level Agent call run in the foreground, explicitly told not to spawn nested sub-agents and to append after each company. Both ran to completion unattended (136 + 194 tool calls). Shard A's detailed per-company findings are in the entry directly above this one (it wrote its own journal note); shard B's summary is folded into the totals below since it didn't write one.

Verification before merge: both shard files validated as clean JSON (23 + 22 lines, 0 bad), all 45 batch company_ids present, zero collisions, zero extras, none already in the master file. enrichment.jsonl 1799 → 1844 (+45). map_roster.mjs: 20061 → 20205 links (+144, lower than the last few batches' ~366-399 — this batch skewed toward single-Gemeinde self-suppliers with short member lists). consolidate_roster.mjs: still 177 shards / 3375 distinct companies (unchanged, expected); roster-links.csv 17079 distinct after dedup. coverage.mjs: population 99.99% → 99.99% (unchanged, expected — batch enriches already-claimed companies), same 24 residual municipalities / 11,023 people, Hartmannsdorf-Reichenau still largest at 1,040.

Non-suppliers found this batch: 0 — all 45 confirmed genuine drinking-water operators.

Field recovery (45 companies, shard A + shard B combined): name/address/phone/website_url 45/45 (shard B: phone 21/22, one gap); email 39/45 found; supply_area_municipalities 44/45; emergency_phone_number 26/45 found (weak field again); contact_form_url only 5/45 found (shard B found literally 0 real forms this pass); address_opening_hours/phone_opening_hours the weakest fields as always (most sites publish only general Rathaus hours, not Wasserwerk-specific ones); handelsregister_nr (shard B, 22 companies): 3 found (real HRB — Pasewalk, Hammelburg, Taufkirchen), 14 not_applicable (KdöR/Eigenbetrieb/eG), 3 absent_from_impressum_page, 2 not_found. analysis_url found for ~37/45; 2 earned an honest not_published after checking full site nav (Kleinblittersdorf, TeWeMa Mandelbachtal); the rest split between not_found (operator's own domain unreachable — Regionalwerke Grafschaft's Cloudflare wall, Gemeinde Malente) and shard B's 4 low-confidence hits with no confirmed own-domain PDF (Pasewalk, Feldolling, Hammelburg, Morsbach). multi_zone true for 15/45 (33%) — below the running ~40-71% baseline, consistent with this batch's tilt toward smaller single-source suppliers.

New host/methodology traps, all measured this pass:

• braunfels.de (Stadtwerke Braunfels, wv-de-15005): HTTP 403 to both WebFetch and curl even with a full browser UA + Accept-Language — the header trick that rescues pirmasens-land.de/ raubling.de does not work here. Durable block, all facts WebSearch-snippet-derived only.
• regionalwerke-grafschaft.de (wv-de-4234): Cloudflare JS challenge page, no fallback content — same class as l.de. Recorded not_found (inconclusive), not not_published.
• wzv-denkendorf.de (wv-de-18867): now dead (502 / TLS UNRECOGNIZED_NAME). Content migrated to wzv-ei.com (shared Landkreis Eichstätt Wasserzweckverbände site, /denkendorf-kipfenberg); wzv-ei.de 301s to .com. Flagging for a domain update rather than changing it inline — domain is the dedup key.
• stadtwerke-hammelburg.de: 403 to everything (WebFetch and curl, all UAs, all paths) — a new durably-unresolvable host in the same class as l.de/braunfels.de.
• kenzingen.de: 403 to WebFetch, 200 to curl with full Chrome UA + Accept-Language; /impressum/ 404s, real path is /rathaus/impressum/.
• rodalben.de: 403 to bare curl (HTML and PDF), needs browser UA.
• gernsheim.de: Wasseranalysedaten PDF list loads via a JS web-component, invisible to static fetch — needs a rendering fetch.
• markt-diedorf.de: one route (/d/wirtschaft/planen-bauen/wasser-und-abwasser) returns HTTP 200 with an empty body — a broken CMS route, not a real 404; don't conclude "not found" from it.

Notable scope/roster findings (not non-suppliers, but worth a human look):

• wv-de-15014 Burgau vs. Stadtwerke Günzburg: both appear to claim Limbach in their supply area per a WebSearch snippet — flagged as a roster conflict, not resolved.
• wv-de-4251 Gemeindewerke Morsbach: good contact details, but no published analysis found despite the TrinkwV obligation, and the 19 served Ortsteile are known only by count, not by name — worth a dedicated follow-up crawl rather than another blind pass.
• wv-de-4230 Wasserbeschaffungsverband Osnabrück-Süd: only 1 of 5 member Wasserwerke (Glandorf) has a public analysis; the other 4 are explicitly "auf Anfrage" only — a genuine publication gap, not a research miss.

For next run: continue with build_enrich_batch_excl.mjs 45. Still 0 pending_domain. New standing items from this pass: Burgau/Günzburg Limbach conflict, Morsbach missing analysis + unnamed Ortsteile, wzv-denkendorf.de→wzv-ei.com domain migration (needs an official_domain update), stadtwerke-hammelburg.de and braunfels.de as new durably-unresolvable hosts. All prior standing items (Beelitz/Meerane roster defects, Solms, Linden/Dahner Felsenland, Glückstadt, Saarwellingen, Gauting/Würmtal, Gilching, Heilbad Heiligenstadt, Ainring/Prenzlau self_only, Oberasbach, Sinzig/Rhein-Ahr, Kronach HRB, Hille/Hartum, Woringen Ungerhausen, Halver/Kierspe, Römerberg-Dudenhofen, Grefrath, Raunheim, WVS Siegen-Wittgenstein, Krumbach/Stauden, Neutraubling, vgwerke-weilerbach.de, OeWA Oebisfelde, Sulz-am-Neckar, Kerken Kamp-Lintfort, Meitingen/ Nordendorf, WLV Ost-Obereichsfeld) still open, none touched this pass.