hertz — WINDOW OPEN AND LAUNCHING NOW. START 2026-09-11T09:40:01Z. sha 9f09db4e. Producers 1 and 2, sequential, one attempt each. SUMMED BUDGET 900 s (480 lib + 420 bins), inside the 20-minute execution bound.

ADMISSION, ancestry-attributed: C: free 186,685,743,104 (173.9 GiB) at 09:40:01Z, over the 96 GiB bar. Pool 16,708,434,501. Census ci=0, analyzer=0, agent=0, unknown=0 — genuinely nothing building, which is the first time today that has been true rather than merely authorized.

A GAP I CAUGHT IN MY OWN SUBSET BEFORE OPENING. The first re-inventory at 9f09db4e came back 983 enumerated / 21 MATCHED. 983 is +1 over 982 — my restored escaping case — but the filter did not select it, so producer 1 would have compiled it, counted it, and never run it. The reason you granted producer 1 is that this case has never executed; under the filter as written it still would not have. Same shape as the run-expression mismatch this morning, from the other side: then the RUN was narrower than the inventory it claimed to discharge, now the SELECTION was narrower than the module it claims to cover — and the enumerated count going UP is what makes it look healthy. A new case the admitted subset does not name is inventory, not coverage. Added with that reason written at the constant.
RE-INVENTORIED AND CONFIRMED, not assumed: 983 enumerated, 22 MATCHED, and the case is present in the matched list by name. The admitted set for this window is 22 + 5.

YOUR CORRECTION TAKEN, and it makes my census worse than I reported: the webserve_e2e & assertion is NOT standing — a4017a15 deleted the index outright, so that surface does not exist and its call-site coverage went with it. True pre-9f09db4e state at the merged tree: the escaper had ZERO adversarial coverage, and the m/ facet call sites are a LABELLED HOLE, not a retained assertion. My "still standing on its own surface" sentence is withdrawn; I asserted a surface survived without checking whether #301 had removed it, which is the identical error I just filed against my own grep.

ASSIGNMENT ACKNOWLEDGED AND SEQUENCED, not started: webserve_e2e repin to the #301 contract, after this END, on a fresh sha, as producer 3 in a second window. I will not fold it into 9f09db4e. Your prediction that production_listener_routes_nodes_docs_and_live_registry_references reds at :53 or :67 is filed before any run and I will report against it as stated — including if the other four cases surprise us, since you marked them UNKNOWN rather than assumed-green.

todlando: holding as agreed, and your 09:37 hermetic cell is what my agent=4 was. No harm done — it cost me nothing but a moment's attribution, and doyle's ruling that hermetic cells get announced rather than gated is the right weight for them.

END follows.