de:wasserversorger-discover

Ran phase-2a batch 5 of the wasserversorger-discover campaign: two synchronous subagents resolved all 60 singleton units (Saarlouis→Altenburg), minting 52 new companies (8 reused via dedupe against existing roster, 1 new pending_domain). Population coverage rose from 59.01% to 61.46% (+2.45pp), municipalities claimed 5,982→6,042; domain-collision check found only the same pre-existing known collision (emkendorf.de), nothing new. Phase-2a pool now stands at 942 units / 15.95M people remaining (next up: Burgdorf); journal updated with next-run instructions, and per SKILL.md's explicit boundary this phase continues unattended until the pool empties or drops below the 10,000-population floor.

Continued phase-2a exactly where the 03:00 run left off: next 60 units of the singleton pool (residual_population >= 10,000, descending) — Saarlouis (37,667) down to Altenburg (31,093). Verified the pool state against the journal before starting (1,002 units / 17.99M people, next-up list matched exactly). Split into 2 synchronous subagents of 30 cities each (shards p2h/p2i), non-overlapping company_id ranges (3225-3254 and 3255-3284), both blocked on in this turn per the skill's synchronous-subagent rule — neither ended the turn early.

Result: 60/60 resolved, 0 unresolved. 52 new companies minted (26 in p2h: wv-de-3225..3250; 26 in p2i: wv-de-3255..3280), 8 reused via the required domain/name dedupe against roster.jsonl before minting (Papenburg→Wasserverband Hümmling wv-de-976, Datteln→GELSENWASSER AG wv-de-122, Ahrensburg→Hamburger Wasserwerke/HAMBURG WASSER wv-de-102, Suhl→ZWAS "Mittlerer Rennsteig" wv-de-636, Geesthacht→Stadtwerke Geesthacht GmbH wv-de-702, Itzehoe→Stadtwerke Itzehoe GmbH wv-de-822, Fürstenwalde/Spree→ZWA Fürstenwalde und Umland wv-de-2602, Oer-Erkenschwick→GELSENWASSER AG wv-de-122 again). 1 new pending_domain row (Merseburger Wasser und Service GmbH, wv-de-3249: no discoverable first-party site, corroborated only via the city's own Beteiligungen page + Handelsregister HRB 215236 Amtsgericht Stendal — recorded confidence: medium).

"Own energy Stadtwerke ≠ water supplier" splits, again the dominant trap pattern: Beckum, Erding, Mühlhausen/Thüringen, Merseburg, Ahrensburg (HAMBURG WASSER, not the local Stadtwerke which is electricity/gas/fibre only), Wermelskirchen (BEW, regional, no city-branded water Stadtwerke), Oer-Erkenschwick (GELSENWASSER direct), Weil am Rhein (retail Stadtwerke but bulk bought from Wasserverband Südliches Markgräflerland), Delbrück and Mörfelden-Walldorf (water run as a direct city Eigenbetrieb, no incorporated GmbH despite a "Stadtwerke"-labelled site nav for the latter), Rheinberg (joint KWW GmbH covering Alpen/Rheinberg/Sonsbeck/parts of Xanten, not city-only), Geestland (Wasserverband Wesermünde, not a city Stadtwerke), Forchheim (three sibling legal entities on one Impressum — water/electricity vs. gas vs. wastewater — had to pick the right one), Hemer (Stadtwerke Hemer remains the legal supplier but leased network operation to ENERVIE Vernetzt GmbH from 2025-01-01). One correction to an earlier roster note: Korschenbroich is served by NEW NiederrheinWasser GmbH, not Kreiswerke Grevenbroich GmbH as a prior batch's note had suggested — re-checked and confirmed on direct refetch before minting.

Decoy/name-collision traps avoided (worth keeping in the SKILL.md list): "Stadtwerke Walldorf" (Baden-Württemberg, unrelated) vs. Mörfelden-Walldorf (Hessen); "Forchheim" and "Kempen" both have unrelated same-named small villages already in the roster from earlier batches (Forchheim am Kaiserstuhl BW, Kempenich RLP) — confirmed via grep against AGS, not just name, before minting. No new host bans/403s beyond the already-documented ones; several sites 404'd on a guessed /impressum path and needed the homepage footer link instead (stadtwerke-kempen.de, stadtwerke-nienburg.de, stadtwerke-wernigerode.de, stadtwerke-emmerich.de, sw-gmhuette.de, plus a handful more) — consistent with the existing caution, nothing new. No CAPTCHAs, IP bans, or forms touched — research only.

Ran the Finish sequence: map_roster.mjs (6,599 links, 0 ambiguous/unmatched) → consolidate_roster.mjs (849 distinct companies, up from 797; 6 shard-level duplicate company_ids found and resolved by last-shard-wins, all 6 pre-existing from earlier batches — none from this run's own shards) → coverage.mjs → build_gv_layer.mjs. Population coverage 59.01% → 61.46% (+2.45pp); municipalities claimed 5,982 → 6,042 (+60, exactly the 60 new links, confirming no accidental double-claims). Domain-collision check across roster.jsonl + companies.jsonl: back to the same 1 known hit (emkendorf.de, unchanged since 2026-08-19) — confirmed clean, nothing new from this batch.

Phase-2a pool remaining: 942 units / 15.95M people (next up: Burgdorf 31,051, Walsrode 31,019, Lohmar 30,992, Bernburg (Saale) 30,992, Weyhe 30,975, Kreuztal 30,905, Bad Hersfeld 30,846, Hoyerswerda 30,759, Rendsburg 30,633, Schönebeck (Elbe) 30,419).

For next run: continue phase-2a from Burgdorf down, same method (2 synchronous subagents, 30 cities each, next new-id range starts at wv-de-3281). Keep doing the dedupe check by name/domain against roster.jsonl before minting — this batch's reuse rate (8/60, plus 1 corrected mislabel) confirms regional suppliers (GELSENWASSER, HAMBURG WASSER, ZWA/ZWAS) keep recurring across neighbouring cities. Pool is now under 16M people and continuing to shrink toward the 10,000- population floor — stop and report per SKILL.md's explicit phase-2b boundary when the pool empties or drops below it, do not descend automatically.