de:wasserversorger-discover

Population coverage rose from 99.21% to 99.40% this run: processed a fresh 60-unit singleton batch via two synchronous subagents, adding 37 companies (3035→3072) and claiming 58 more municipalities (10399→10457, residual now 486/504.8k people), with 2 genuine unresolved (Rieden, Sauldorf). Both shards self-caught a duplicate mint via the pre-merge domain-collision check before it reached the roster, so the standing collision backlog stayed flat at 29 with zero new ones introduced; journal updated with two new data-quality flags (a swapped AGS pair, a mismatched hq_ags) for the still-overdue human consolidation pass.

Followed the skill's synchronous-shard protocol: two general-purpose subagents dispatched with run_in_background: false and the mandatory "no Agent/Task tool" line. Both completed inline within this turn (durations 878s/842s), neither recursed into background sub-subagents.

State on entry: picked up after the 20:30 wasserversorger-enrich run, landed on disk at 99.21%/10399 claimed/544 residual, 3035 companies. Found a stale leftover batch60-input.json (mtime 20:34, filed under "never trust leftover files") — checked it against a fresh coverage.mjs/build_gv_layer.mjs run first: 59 of its 60 AGS were already claimed (the same batch the 18:34 run had already consumed and merged), so discarded it and generated a genuinely fresh batch with build_batch60.mjs rather than reusing it.

Batch: build_batch60.mjs — 60 singletons, residual_population descending. Split 30/30 into two synchronous subagents, company_id ranges wv-de-19510–19539 (shard A) / wv-de-19540–19569 (shard B).

Shard A (30): 16 new mints, 13 reused existing companies, 1 unresolved (Rieden, Ostallgäu — checked own site, a 404'd Behördenwegweiser page, no naming Zweckverband, and a non-machine- extractable Landkreis Wasserschutzgebiet PDF; recorded as an honest gap). Self-caught one duplicate mint before merge: wv-de-19523 ("Erzgebirge Trinkwasser GmbH") duplicated the existing wv-de-3857 (verified via Handelsregister HRB 11896) — deleted and re-pointed the Frauenstein link to the existing ID. Two name-collision traps avoided (Burghaslach vs. the unrelated Burgsalach; Hessen's Friedewald vs. an unrelated Rheinland-Pfalz Friedewald already in the roster). Flagged without fixing: a pre-existing roster-links.jsonl row has St. Peter/St. Märgen AGS swapped (08315094/08315095) — not this shard's file to edit, left as a note for the standing data-quality backlog.

Shard B (30): 21 new mints (mostly small-Gemeinde self-supply), 8 reused existing companies, 1 unresolved (Sauldorf, Baden-Württemberg — re-checked prior research plus fresh search, still no verifiable supplier). Self-caught one duplicate mint before merge: a new "Markt Kastl" row (kastl.de) duplicated the existing wv-de-10532 for the same Markt Kastl im Lauterachtal — deleted and re-pointed, wv-de-19550 permanently retired/unused. Flagged without fixing: existing wv-de-10532.hq_ags is recorded as 09377128, which doesn't match this municipality's real AGS 09371132 — noted in the link row for a future data-quality pass.

Validation before merge: both shards well-formed JSONL (16/16, 21/21 roster rows; 30/30, 30/30 link rows, 0 malformed). AGS-coverage check against batch60-input.json: all 60 assigned AGS present across the combined link rows exactly once, 0 missing, 0 duplicated (58 resolved + 2 explicit unresolved = Rieden + Sauldorf). Cross-shard + full-roster eTLD+1 domain-collision recheck run before consolidate_roster.mjs, per the skill's rule: 29 standing pre-existing collisions, 0 touching either new shard.

Merge: map_roster.mjs — needs-review queue unchanged at 1231 (AGS-seeded singleton rows don't feed it, as expected; total link rows 14643 → 14703, +60 exact). consolidate_roster.mjs: 3035 → 3072 distinct companies (+37 = 16+21, exact). coverage.mjs: municipalities claimed 10399 → 10457 (+58, exact — 2 of 60 stayed unresolved). Population 99.21% → 99.4% (+0.19pp), residual 544 → 486 municipalities / 656k → 505k people. build_gv_layer.mjs: multi-member pool unchanged at 10 units/22 municipalities (still on hold), singletons 522 → 464 (-58, exact). Post-merge full-roster domain-collision recheck (eTLD+1): still 29, confirming this run introduced none.

Crossed 99.4% population coverage.

For next run: continue singleton batches by residual_population descending — 464 singletons remain, largest still ~6.6k (Schelklingen etc., same-day-excluded until the date rolls over past 2026-08-31 — no manual handling needed). Check for and discard any stale leftover batch60-* files against a fresh coverage run before reusing them — this run's leftover was 98% already consumed. The 10-unit/22-municipality multi-member pool remains on hold pending the overdue human decision — do not spend a run on it without new information. Standing domain-collision backlog unchanged at 29 (eTLD+1) — still awaiting the overdue human consolidation pass, along with the 6 standing duplicate-mint name pairs from prior entries (unchanged, not touched this run). New data-quality flags for a human pass: a swapped St. Peter/St. Märgen AGS pair in roster-links.jsonl (08315094/08315095), and wv-de-10532's hq_ags (09377128) not matching Markt Kastl's real AGS (09371132). _tasks/netzbetreiber-contact/german-water/ remains absent from this checkout (noted every run, out of scope for discovery).