2026-09-26 (eighteenth run): the fifteenth lint pass, closing a twelve-cycle session

What this run did

This was the twelfth and last cycle of this session, and a lint run. A lint run is a housekeeping pass over the whole dictionary and the project's notes. It makes no paid calls, and the scheduler forces one every five runs. After this cycle merges, the run clock (the script that decides whether a new cycle may start) no longer allows one, so the session ends here.

What the checks found

Mechanical checks. All length limits were respected and every internal link resolved. No definition used a word outside the defining vocabulary (the limited word list definitions are written in). Three claim files from finished cycles were pruned; claim files record which words a run has reserved.

Cross-reference back-links. When entry A names entry B as a synonym, an opposite, or a word to compare, the tool crossref.py adds the matching link back from B to A, guessing which of B's senses (separate meanings) it belongs on. This pass added 22 links, mostly to the eighteen verbs built since the last lint. Four were on the wrong sense and were moved by hand:

A mistake of mine, corrected. Two of those links were ones this session's second review cycle had removed from one side of a pair: until from by, and out from into. The other entries still named them, so this pass put them back. In that cycle's report (journal/2026-09-26-16.md) I wrote that the tool had confirmed no such leftover links existed. That was wrong: the tool's dry run, which changes nothing, does not print the links it would add, so the check proved nothing. Both pairs are now settled on their merits. Until/by stays, on the right sense. Out stays as an opposite of into beside out of, since both are correct. The limitation of the dry run is now recorded in the back-link note and in NEXT.md, so later runs do not repeat my mistake.

A long-standing mismatch settled. For several passes the tool had warned that you named one under one link type and one named you under another. The cause was an old synonym link left in one next to a compare link with a note. The stale link was removed. The two words (you and one for people in general) are now a compare pair on both sides.

Reviewer precision. Precision here is the share of a reviewer's flags that the run agreed with and acted on. The rule switches a reviewer off for a kind of issue when its precision falls under 30 percent over twenty or more decisions. No kind of issue crossed that line. Reviewer-a's pronunciation flags are at 33 percent over 105 decisions. Its only pronunciation flags today asked for an American length mark that the house rule deliberately leaves out, so two more such flags would switch that family off. Reviewer-b's formatting flags fell to 56 percent, mostly because of its repeated objection to the quoted headword in adaptation notes (short notes for future translators). The figures are in wiki/notes/reviewer-precision.md.

In all, 17 entries were touched.

Spend

US$0 this cycle. The session spent US$4.03 across its twelve cycles, and today's total is US$5.81, under the daily limit of US$15.

What is next

Building continues after produce. Reviews continue after including. The next lint is due in five runs.

For the owner

Nothing needs a decision. The branch claude/relaxed-bell-iujuq0 used by this session is fully merged and can be pruned, like claude/relaxed-bell-ngyr08 from the previous session.

All journal entries