Run report, 2026-11-11, second run (run 20261111T084557Z-ex1wz5)

Cycle 1: review

What this cycle did. The selector (the tool that picks each cycle's job from what is most overdue) chose review: a second reading of entries that are already published, to find what they are missing. There are no draft entries waiting and no open "Entry report" issues from readers, so the cycle went on with the coverage pass, the owner's ruling B1 of 2026-10-24: the most important band-1 words (band 1 is the commonest group of words) are read one by one, those with the most senses first, and anything a learner would expect to find but cannot is added. Four entries were read: pick (verb), picture, play, and pot (nouns). The pass has now covered 287 entries; the next is program.

Before the work. No open pull requests from an earlier run, nothing in inbox/ (the folder where you leave instructions), no reader reports. The old work branch claude/friendly-mayer-mdy32d still holds nothing that is not in main and already has its line in reviews/needs_curator.txt asking for deletion.

What was added.

What the reviewers found and what was decided. Two reviewer models read each changed entry field by field. They raised 47 issues, 5 of them "blocking" (must be settled before the entry stays published). 40 were accepted and fixed, 7 rejected.

Spend. US$0.25 for this cycle; US$0.98 of the day's US$15 so far.

What did not work. Nothing failed. Two idioms noticed on the way, pick your way and pick holes in something, belong under way and hole, which do not yet have them; they are listed in NEXT.md for those entries' next review.

What is next. More cycles of this run, chosen by the selector. The coverage pass continues with program.

What needs you. Nothing new from this cycle.

Cycle 2: build

What this cycle did. The selector chose build: writing new entries from the queue (the ordered list of words still to be written). The cap for a cycle is twelve; this one wrote eight, so that each could be full: lighting, lion, listener, litter, living, lizard, local, and location, all nouns. The dictionary now has 4,176 entries, all marked reviewed.

The entries. A few points worth knowing:

What the reviewers found and what was decided. Both reviewer models read every field of every entry. They raised 47 issues, 16 of them blocking (must be settled before an entry is published). 43 were accepted and fixed; 4 were rejected.

Spend. US$0.24 for this cycle; US$1.22 of the day's US$15 so far.

What did not work. Nothing failed. Two connection errors to the second reviewer's service were retried automatically and succeeded.

What is next. Further cycles as the selector decides; the next build starts after location in queue order.

What needs you. Nothing new from this cycle.

Cycle 3: closure

What this cycle did. The selector chose closure: writing entries for words the dictionary already uses but does not yet define, so that a reader who meets a word in a definition can look it up. Words come first from definitions (the "closure" source), then from word families (locker from lock), then from cross-references. Six entries were written: senior, locker, lover, lunchtime, majority, and recording, all nouns. The dictionary now has 4,182 entries, all reviewed.

A queue fix. The closure source had queued recordings (from a definition of label), a plural form rather than a dictionary word. It was marked as a duplicate and the singular recording was written instead.

The entries.

What the reviewers found and what was decided. 20 issues, 5 blocking; 19 accepted, 1 rejected.

Spend. US$0.17 for this cycle; US$1.38 of the day's US$15 so far.

What did not work. The closure source queued a plural form (recordings) as if it were a headword. This is a known weakness of the tool that reads definition words; it was handled by hand and needs no new machinery yet.

What is next. The run may start a fourth cycle; family rows continue after majority.

What needs you. Nothing new from this cycle.

All journal entries