Yezur Wiki:Daily routine/Report

Working report of the Alompan daily routine (how it works). Every finding lives as one row below until it is resolved: set a row's Status to give a verdict, and the next run carries it out and clears the row. Nothing on the wiki is changed except what is explicitly approved or amended.

The possible statuses:

pending — awaiting review; no action is taken (the default)
approved or accepted — carry out the Suggested fix, as written
amended — carry out the fix described in Notes instead
deferred — real, but not now; moves to Dismissed flags and is not re-flagged
rejected — not an issue; moves to Dismissed flags and is not re-flagged
to to-do — real issue that should be marked on User talk:Yezur's to-do list instead for human intervention
fixed — real issue, but was a simple fix for Yezur, or has by other means already been taken care of; confirm and clear

To revive a dismissed flag, move it back into the table as pending.

Flagged items

#	Flagged	Type	Severity	Page(s)	Issue	Suggested fix	Status	Notes
72	2026-08-04	structural	medium	270 `Dictionary:` entries (Proto-Unterian)	*Every Proto-Unterian entry states where it came from in prose, and not one of them says so to Lexemes.* The 270 entries written this window each carry an `===Etymology===` section naming their source — 257 of the form "From Proto-Andusian `{{m\|YCP-XX\|…}}`" and 13 of the form "A Proto-Andusian-age formation, X + Y, continued only in Unterian". None passes `etym_type` or `etym_source` to its `{{head}}` call, so all 270 rows reach the Lexemes table with both fields empty. Yezur Wiki:Conventions gives that table one job — "cognate sets, borrowing chains, … one query each" — and the query it exists for, every Proto-Unterian word inherited from Proto-Andusian, currently returns nothing, on a language where the answer is 270. Ancient Hertic is the comparison: 209 of its 212 rows carry both fields, and the three that do not carry `{{etym-missing}}` instead and sit correctly in Category:Ancient Hertic entries needing etymology — that language files its filiation, this one does not. The other four languages with entries are root protos or single lemmas, so Proto-Unterian is the wiki's only descendant language whose descent is unrecorded in structured data.	Mechanical, and asserts nothing the entries do not already say. Add `\|etym_type=inherited\|etym_source=YCP-XX` to the `{{head}}` call of the 257 entries whose etymology reads "From Proto-Andusian …", and `\|etym_type=compound\|etym_source=YCP-XX` to the 13 formations — the same two values already in use in the table (inherited 209, compound 1). The source is read from each page's own etymology line, not inferred, and no definition, gloss or heading changes. One pass; the routine can do it the day it is approved.	pending	New 2026-08-04, from the Lexemes check. Not a criticism of the entries: the prose etymologies are complete and consistent, and this is only the machine-readable half of what they already say. Recorded as not a finding beside it: the 273 Proto-Andusian, 211 Proto-Herta-Olaic and 102 Proto-Njeshan rows also carry no etymology, and are right not to — they are the reconstructed roots of their families and have no source on the wiki to name. The three Andusian branch protos registered this window (`YCP-XA`, `YCP-XB`) have no entries yet; when they do, they will be the same case as this row.
73	2026-08-04	structural	low	Yezur Wiki:Maps	The one page in the project namespace that is not a project page. Asked to check whether every `Yezur Wiki:` page belongs there (see #47, dismissed), the routine read all six. Five are policy or project documentation and sit correctly: Yezur Wiki:Conventions, Yezur Wiki:Daily routine and its Report, Yezur Wiki:Dictionary structure, and Yezur Wiki:Main Page. The sixth, Yezur Wiki:Maps, is a software smoke-test for the YezurMaps extension: it opens "YezurMaps extension test" and "If you can read this but see no maps below, the module didn't attach — check `Special:Version` lists 'YezurMaps'", then runs four `{{#map:}}` calls to prove the parser function fires, that both tile backends are reachable, and that Overpass geometry loads. It addresses the reader directly throughout ("Replace 42 with a REAL relation ID from your data", "a route you've actually mapped") and its content is placeholder test data — First test province, Test marker with popup, relation ids 42 and 148–150. It is also the only page in the namespace that is uncategorised and the only one that nothing on the wiki links to.	flag-only. Where a live extension test belongs is a decision the convention set reserves to the operator, and it is the concrete case #47 was really about: if `Meta:` is to hold anything, this page is the readiest candidate, and moving it would give that namespace its first page and the sidebar's Project docs link somewhere to lead. Three answers are all defensible — move it to `Meta:`, leave it where it is and categorise it, or delete it now that the extension is known to work. The routine holds the `move` right and can carry out whichever is chosen.	pending	New 2026-08-04, and the direct product of #47(amended), which asked exactly this question. Recorded as not a finding: Yezur Wiki:Main Page is uncategorised too, but that is #14 (deferred), which covers the portal front pages as a class.
74	2026-08-04	naming	low	Template:Infobox polity, Yezur Wiki:Conventions	The house word was settled this run, and the one template that uses the other word is the one nothing uses. Carrying out #51 recorded the operator's ruling that country is the house word in preference to polity, and rewrote both worked examples on Yezur Wiki:Conventions to read "Countries in Andusia". Template:Infobox polity is now the only place on the wiki still carrying the retired word — and it is one of the three entries on Special:UnusedTemplates, transcluded by no page at all. Category:Countries holds one member (Engil) and already uses the settled word; Template:Infobox former country and Category:Former countries already agree with it too. The Polities Cargo table the template once declared no longer exists — it was removed with the rest of the encyclopedia's Cargo (see #43, dismissed) — so no data anywhere is keyed to the name.	Cheap, and cheapest now: rename Template:Infobox polity to Template:Infobox country, leaving a redirect, and update the one word in its documentation. Nothing transcludes it, so nothing can break, and the routine holds the `move` right. The alternative is equally clean: say that polity is wanted as a distinct term — a living state as against a country in the broader sense — in which case the template is correctly named and the row closes with a sentence on Yezur Wiki:Conventions saying so. What is not wanted is the present state, where the convention page names one word and the only template on the subject names the other.	pending	New 2026-08-04, and the residue of an operator ruling rather than a fresh defect. The ruling settled the worked examples, which is what #51 asked; this is the consequence it left behind. #43 (rejected) is not re-opened — that row was about the Cargo table, which is gone; this is about the name on the template.
75	2026-08-04	structural	low	Herta-Olaic, Njeshan, Tacchoan, List of languages	Two hand-maintained lists now write the same twelve names two different ways, because two verdicts said so. #67 was approved as written, and its fix says the missing Members rows go in "as plain text where no article exists … so no red link is created". #71 was amended in the same triage, and its note says the opposite for the roster — "red-link the rest: link all twelve roster entries". Both were carried out this run, exactly as instructed. The result is that Airananwe, Anilonic, Senitian, Usinger, Esselitian, High Gaillean, Kankai, Middle Airananwe, Old Ousperian, Old Usinger, Skennic and Terjiquan are blue-linked on List of languages and plain text in the Members table of the family article that lists them. Every other name in those same tables is linked. The two lists agree perfectly on membership — the check confirms 6/6, 8/8 and 7/7 against the roster — and disagree only on this.	Mechanical either way, and one pass. (a) Link them in the Members tables too, in the `[[X language\|X]]` form the roster now uses — matching every other row of those tables and making the two lists identical in form as well as content. This adds no new red-link targets: all twelve are already wanted from the roster. (b) Or unlink them on the roster, if the red links there were meant to be temporary. (a) is the reading that follows from #71, which is the later verdict.	pending	New 2026-08-04. Filed rather than assumed: both instructions were explicit and scoped to different pages, so the routine carried out each as given rather than harmonising them on its own. Not the same as #57 (deferred), which is about whether the four Tacchoan branch names get articles — these twelve are languages, and the roster already treats them as wanted.
76	2026-08-04	factual	low	Dictionary:kidz, Dictionary:kizi, Dictionary:oʊmjiəɲɔ	*Three entries call the peoples of Yezur human, and nothing on the wiki says whether that is the word.* The three are the only in-frame uses of human anywhere on the wiki, checked across all 1,264 pages: "# person, human being" on the Proto-Andusian entry and on its Proto-Unterian descendant, and "# a human being, someone; one of us as against a beast" on the Proto-Herta-Olaic one. Every other occurrence of the word on the wiki is out-of-frame, describing the human editor. The three are consistent with each other and read naturally; the question is only whether the encyclopedia's own voice uses human for Yezuri people, which touches Yezur Wiki:Conventions §The Earth test — the same section that rules on "Yezuri" as a reflexive qualifier — and would settle how every future entry glossing person is written.	flag-only. A canon question, not a maintenance one: the routine will not choose a word for the peoples of the world. Say whether human is in-frame vocabulary. If it is, the row closes and nothing changes. If it is not, the fix is three definition lines and the routine can make it in one pass.	pending	New 2026-08-04 as a row, though not a new observation: it was recorded inside #62's notes on 2026-07-30, when there was one such entry. That row was cleared this run, so the question would otherwise have gone with it; it is filed here so it has a home of its own. It has also grown from one entry to three, all three by inheritance from the same gloss.
77	2026-08-04	structural	low	Dictionary:About Proto-Andusian, Dictionary:About Proto-North-Andusian, Dictionary:About Proto-South-Andusian, Dictionary:About Proto-Unterian	The Dictionary has grown a class of appendix page and no category for it. Four `Dictionary:About …` pages now document a proto-language apiece — its conventions, phonology, sound laws from its parent, derivation and reduplication. They are the only Dictionary pages that are not entries, they are written to a common shape, they cross-link one another, and all four are uncategorised: they are four of the seven entries on Special:UncategorizedPages. No category on the wiki fits them — Category:Dictionary maintenance is for review queues and is hidden, and the `{{head}}`-emitted lemma categories only apply to entries. More will follow: three further proto-languages are registered in Module:Languages and two of them have no About page yet.	One category, then four one-line edits — but the name and placement are the operator's call, as Yezur Wiki:Conventions reserves new categories. The shape is plain: a category for Dictionary pages that describe a language rather than a word, parented to Category:Languages or to a Dictionary branch of it. Once named, the routine can create it and file the four.	pending	New 2026-08-04. Distinct from #14 (deferred), which covers the three portal front pages and the root Category:Yezur — the other three names on the same list. Recorded as not' a finding: these four pages are also on Special:LonelyPages and Special:DeadEndPages by the same measure as every Dictionary entry.
78	2026-08-04	structural	low	File:Tacchoan-subfamilies.png, Tacchoan	The file's description is the article's caption, copied, and the caption has just changed. File:Tacchoan-subfamilies.png's description page was the Tacchoan §Members caption word for word. Carrying out #55(a) rewrote that caption this run — it now describes what the map shows and names Herta-Olaic as the neighbouring family rather than a branch — while the description page still carries the old wording, because nothing keeps the two in step. #54 asked three questions about file description pages; the operator answered two of them (files are categorised; a source line where one is needed), and this is the third, left open: whether a description page may restate an article caption or should point at it. The wiki has one file today, so the answer costs one edit; #53 (deferred) means a corrected export is coming, which will make it two.	Mechanical once decided, and the decision is small. (a) Point rather than restate: the description page says what the file is and where it is used, and leaves the colour key to the article. (b) Or restate, and accept that the two are kept in step by hand — in which case the routine can copy the new caption across now and check the pair on every run. Either way the file's categories, added this run, stay as they are.	pending	New 2026-08-04. Not a re-opening of #54, which was carried out as amended and cleared: this is the one question the amendment did not reach, filed so it is not lost. Not a re-opening of #53 either — that row is about the image itself, this is about the page describing it.

Dismissed flags

Deferred or rejected findings, kept here so later runs do not re-flag them.

#8 (deferred) — Wanted history-period pages Yezuri Age of Exploration / Yezuri Middle Ages. Operator: not canon names for these eras; will not create stubs under temporary names.
#11 (deferred) — Andusian / Tacchoan / Shung-Worrehan member-language queries fall back to the default. A content gap, not a data error; operator: no member languages are in yet, add when ready.
#12 (rejected) — Greek-letter native names on Hertia / Hertic language / Alompe etc. Operator: fully intentional in-world orthographic style.
#14 (deferred) — Uncategorised portal front pages and the root Category:Yezur. Operator: accepted, but no decided approach for handling it yet.
#21 (deferred) — The four wanted Ancient Hertic inflection templates (conj-IS, conj-OO, decl-N1, decl-N2). Operator: "will be fixed when the tables are created — already on my to-do." Confirmed present on the to-do list at User talk:Yezur.
#30 (rejected) — The one-line User:Yezur page, read as an unfinished sentence. Operator: the line is not unfinished — the city name is written there as , "using a script I have added to the wiki, with characters stored in the PUA … a temporary text intended to be used for testing the PUA font; it is working correctly." Answering the operator's condition for reflagging: the routine can see the characters and reads five PUA codepoints there (U+E067 U+E111 U+E179 U+E179 U+E0EB, the third and fourth identical, so a five-letter name with a doubled letter), which round-trip intact through the API. What it cannot do is read them: no mapping from those codepoints to sounds or letters exists anywhere on the wiki, so the glyphs are legible only to a reader with the font. Not reflagged. If a transliteration table for the script ever lands on-wiki, say so and later runs can check such names instead of skipping them.
#32 (rejected) — Airananwe and the code name:yas-ai in Map:name's worked example matching no language and no registry code. Operator: "that language will land later. No need right now." Left as written; the routine will not re-open it when the language arrives.
#33 (deferred) — The two redlinked values Map:place=hamlet and Map:place=suburb in Map:place's settlement table. Real but not now; they remain the only redlinked values in any key page's value table, and both are listed in the new Map:Features index only through their parent key.
#43 (rejected) — Template:Infobox former country declared and stored into a FormerCountries Cargo table that was never created, so Hertia's structured data went nowhere. Operator: "Cargo tables no longer the go-to." Overtaken by the retirement of Cargo from the encyclopedia (2026-07-30): the declaration and the store have both been removed from the template, so there is no longer a table to create. Confirmed this run — of the six tables the wiki once declared, only Lexemes answers a query; Continents, Languages, LanguageFamilies, Polities and FormerCountries now all throw the same MWException that once distinguished an absent table from an empty one, because all five are now absent. The administrative Create data click this row was waiting on is no longer wanted.
#59 (rejected) — Betonia, Unteria and Sea of Sihia categorised but reaching no Cargo table, with no infobox on the wiki fitting a subcontinent, an island or a sea. Operator: "No longer relevant since Cargo tables no longer the go-to. Not everything needs to be tabulized." The structured-data half of the finding has been removed by the same ruling, and the infobox question closes with it. Vetsihia's borrowed {{Infobox continent}} is left as it stands: with Cargo gone it is display only, and its type=Geographic region (colloquial) already says what the article is.
#47 (deferred) — Meta: and Help: documented as live namespaces while holding no page, and the sidebar's Project docs link leading to an empty index. Operator: "scarcely used, but will become hugely useful should the project ever open to others — there is simply nothing yet to meta about on a large scale," and Help: is in the same position; no edits to MediaWiki:Sidebar, Yezur Wiki:Main Page or Yezur Wiki:Conventions. The amendment's own task — check whether any Yezur Wiki: page could move to Meta: — was carried out this run, and its one finding is filed as #73. Standing consequence, recorded rather than re-flagged: the sidebar link every reader sees still leads to an empty index, by decision.
#50 (rejected) — The unreachable index-only branch in Engine:handle_affix of Module:Auphen, where the ELS ~= guard makes the second branch untestable. Operator: "let's not mess with the module's code." The routine will not re-open it; if an affix rule with an index-only else clause is ever written and silently does nothing, that is this row.
#53 (deferred) — The OpenStreetMap attribution strip baked into File:Tacchoan-subfamilies.png, and the colour key that gives yellow to two branches. Operator: "the operator will upload a better export in time." The licence question is closed: the map is not OSM-derived — it only uses the OSM software — so no credit is required and the strip can simply go. Part (c), the caption clause, was carried out under #55 this run.
#57 (deferred) — The four Tacchoan branch names (Esselis, Rumesselis, Kipoa, Usy) and the island Usings named into existence with no article, and linked in the caption but plain in the bullets that define them. Operator: deferred. The five remain the only red links out of Tacchoan; the branch names stay plain in the bullets.
#58 (deferred) — The five Talk:Family A–Talk:Family E pages queued for deletion since 2026-07-28. Operator: "the operator will get to the deletions in due time." Listed under Deletion queue each run as the task list requires, but not re-flagged.
#65 (deferred) — Map:Conventions, promised by Map:Main Page's lead and linked from two pages, never written. Operator: deferred. It remains the only red link in the Map: namespace outside dismissed #33.
#70 (deferred) — The 37 superseded Proto-Andusian entries at implosive titles, queued for deletion by the operator's own ruling of 2026-08-03. Operator: "as #58, the operator will get to the deletions in due time." Listed under Deletion queue each run, not re-flagged.

Latest run

2026-08-04. The run the previous five were waiting for. Every one of the 22 rows that had stood pending — some since 2026-07-26 — carried a verdict when this run opened, left in one triage by the operator through Thiorosan (rev 2788) and extended twice more during the day. Phase 1 has never had work before; this time it had all of it. 90 edits were made under those verdicts, and the table is empty of old rows for the first time since the routine began.

The survey then read a second large window: 955 changes on 653 pages, 234 of them new, made between 19:54 on 2026-08-03 and 13:40 on 2026-08-04 — Thiorosan 951, Khurouan 3. Almost all of it is Dictionary work, and its centre is a new language.

What the operator's verdicts came to

Of the 22: 12 carried out as edits (7 approved as written, 5 amended), 3 confirmed as already fixed and cleared, and 7 dismissed — one rejected and six deferred, the sixth being #47, which deferred itself once the audit it asked for had been done. Every row is off the table. Three of the amendments — #51, #54, #69 — changed the shape of the fix rather than its existence, and are described below where they land.

The largest pass was #63: 77 definition lines on 72 pages rewritten from a plain parenthetical to a label call, # (by extension) X → # {{lb|YAP-XX|by extension}} X. The measured scope had grown since the row was filed — 75 Proto-Herta-Olaic lines as recorded, plus two (of highland water) lines rather than one: the Proto-Andusian entry the row already named, and a Proto-Unterian entry that inherited the same line when the language was seeded this window. Both were folded in, at their own languages' codes. Checked before and after: none of the five labels is in Module:Labels' categorising vocabulary, so the pass added no category to any page — verified on all 72 afterwards, where the only categories are the {{head}}-emitted lemma and part-of-speech ones, and the labels now render italic as intended.

The widest pass was #66(a): 19 background declarations across 7 pages given the explicit color:#202122 they were relying on the skin to supply — 15 inline on the four portals and Template:Delete, and 4 in the two TemplateStyles sheets (Template:Infobox/styles.css, Template:Translations/styles.css), which the row's corrected count had found last run. Option (b) was not taken up, so the sheets were fixed in place rather than absorbed. Nothing renders differently today; the night-mode switch is now safe to flip. Re-scanned afterwards: zero background-without-colour declarations remain anywhere on the wiki.

The rest, page by page. Yezur Wiki:Conventions took three rows in one edit — #51 (both worked examples now read "Countries in Andusia", settling both the connector and the house word), #52 (a §Maps bullet for {{On the map}}, restating only that template's own usage note) and #54 (a new §Files recording the operator's answers: files are categorised, and a source line is written where one is needed). #54 also created Category:Files at the top level with Category:Linguistic maps and Category:Maps of Taccho beneath it, and filed the wiki's one file into the latter two — Special:UncategorizedImages is now empty. Tacchoan took three rows in one edit — #55 (four wordings), #56 (Usy is no longer called an isolate: it is now "the family's only single-language branch … it stands apart from the other three branches while remaining within the family", which is the operator's sentence rather than the routine's suggested wording) and #67 (the seven-row Members table replacing "No member languages are currently listed."). #67 also added four rows to Herta-Olaic and one to Njeshan. #71 linked eleven roster languages and three plain family names on List of languages, and two more language names in Notes cells that the rest of the column already linked. #61(a) narrowed Yezur Wiki:Main Page's description of Cargo to the Dictionary's Lexemes table and dropped it from the prefer templates and Lua bullet. #68 brought Yezur Wiki:Dictionary structure up to the Dictionary it describes, in three places. #69 replaced the nine placeholder words on Template:Translations/testcases with real lemmas.

The three rows marked fixed, confirmed

Each was re-read rather than taken on trust.

#60 — the high-severity Map: breakage. siteinfo now reports namespace 3004 and 3005 as case-sensitive again; all 33 lowercase-titled pages resolve by name with no normalisation and no redirect; Map:Features renders 34 of its 35 links blue, the 35th being Map:Conventions, which is deferred #65. Fixed at the source, in configuration, exactly as fix (a) asked.
#62 and #64 — the Dictionary separator and quoting passes, carried out by Thiorosan on the operator's direct instruction: 163 definition lines on 158 pages. Confirmed independently across all 2,190 definition lines on the wiki: zero straight-quoted glosses remain, in any language. The ten lines still carrying two or more semicolons are all on the Dictionary:About … grammar pages, where they list sound changes rather than senses, and are correctly outside the ruling. The convention itself — commas join glosses, a semicolon introduces explanation — is now written down at Yezur Wiki:Dictionary structure, which is what makes the pass checkable at all.

What the window contained

A sixth language, and the first descendant proto. Proto-Unterian (YCP-XC) was registered with its two sister branch protos — Proto-North-Andusian (YCP-XA) and Proto-South-Andusian (YCP-XB) — and then seeded with 270 entries, every one of them derived from a Proto-Andusian ancestor and carrying a prose etymology naming it. Andusian gained the three-branch structure; Dictionary:About Proto-North-Andusian, Dictionary:About Proto-South-Andusian and Dictionary:About Proto-Unterian document each branch's sound laws. It is the wiki's second-largest language on the day it was created, and it arrives through another language rather than beside it — the first time the Dictionary has recorded descent at scale.

Descendants lines became machinery. Template:Desc and a desc() function in Module:Languages were built so a Descendants bullet resolves its language name from the registry, and 269 entries were converted to it. The stated reason is in the edit summary — "so language renames never touch entry pages" — and it is the same principle the Members tables were converted away from and the translations box was built on: the registry names things once.

185 etymologies were trimmed to bare filiation on operator request, and 36 Proto-Andusian entries had reconstruction hedges dropped in a further register pass.

The finding this window produced

#72, and it is the mirror image of the good work above. All 270 Proto-Unterian entries say in prose where they came from; none of them says so to Lexemes. The etym_type and etym_source fields are empty on every one of the 270 rows, so the query the table exists to answer — every Proto-Unterian word inherited from Proto-Andusian — returns nothing, on the one language where the answer is the whole vocabulary. Ancient Hertic files those fields on 209 of its 212 rows, and the three exceptions carry {{etym-missing}} and sit in the review category built for them, so the wiki already has the habit; this language did not inherit it. The fix reads each page's own etymology line and adds two parameters to its {{head}} call — 257 inherited, 13 compound, both values already in the table.

Six further rows are filed, all low. #73 is the Yezur Wiki:Maps finding that #47(amended) asked for by name. #74, #75, #76 and #78 are each the residue of a verdict carried out this run — a ruling's unapplied consequence, two instructions that scoped differently, and two questions that were living inside rows now cleared. #77 is the four uncategorised Dictionary:About … pages.

The content review

Clean, and the scan was wide. Across all 650 changed in-frame pages, read with HTML comments stripped: three Earth-test hits, all the word human in a gloss of person, which is #76 — and, checked across all 1,264 pages, the only in-frame uses of the word anywhere. Forty-one apparent voice hits, every one of them a false positive on inspection: they are the pronouns I, you and we glossing pronoun entries — words as words, not address to a reader. No break of frame, no design-process reference, no chatty aside in any of the 233 new entries.

Structurally the new entries hold. All 233 carry a {{head}} call, an L2 language heading, an L3 part-of-speech section and at least one numbered definition — except the three Dictionary:About … pages, which are appendix pages rather than entries and correctly have none of that. The 270 Proto-Unterian etymology lines fall into exactly two shapes, and all 270 name Proto-Andusian; there is no third case for #72 to mishandle.

The lists, checked end to end

The hand-maintained-lists check, which fired for the first time last run, is clean again — and clean because it was fixed, not because it stopped looking.

Every family Members table now matches the roster exactly: Herta-Olaic 6 of 6, Njeshan 8 of 8, Tacchoan 7 of 7, with Andusian and Shung-Worrehan correctly empty. Nothing missing, nothing extra.
All 23 roster codes resolve in Module:Languages and agree with it on name and family. The registry's 29 codes exceed the roster's 23 by exactly the six reconstructed protos, which the roster's own lead excludes.
All 10 {{Infobox language}} articles agree with the registry on family; English language carries two codes and no family, correctly, being an isolate.
Main Page still has no continent list — it links Category:Continents, which holds 7 — so there is nothing there to drift.

Clean

Linter 0, over every category the extension reports, and corroborated by hand across all 1,264 pages. Broken redirects 0. Double redirects 0. Auphen errors 0 across the window's transclusions, and test failures 0. Unused images 0. Uncategorised images 0 — that one cleared this run by #54.

Lexemes is 1,070 rows (was 800): 0 missing a key field, 0 with an empty gloss, 0 duplicate (word, lang, pos), 0 rows pointing at a page that does not exist, and word equal to the page title on 1,070 of 1,070. 42 pages now file more than one row — the Proto-Andusian and Proto-Unterian sections of one shared title, or two parts of speech on one page — which is the form the blueprint prescribes, so multiple rows per page is correct and only the triple is tested for uniqueness. Etymology fields are filled on 210 rows and half-filled on none.

The counts that moved, and why. Wanted pages 43 → 319: 275 of them are Dictionary entries wanted by the new Descendants lines, which name the reflexes of the three Andusian branches before those branches have entries — a deliberate consequence of the window's work, not a defect. Twelve more are the language articles #71 red-linked on the operator's instruction. Orphaned 298 → 164 and dead-end 289 → 122, both falling for the same reason: the Descendants lines interlink entries that stood alone before. Wanted categories 59 → 69, every one of them {{head}}- or {{lb}}-emitted and all covered by to-do #18; the 10 new ones are Proto-Unterian's. Wanted templates 4, unchanged, and all four are deferred #21. Uncategorised pages 4 → 7, the three additions being the new Dictionary:About … pages of #77.

Deletion queue: 42, unchanged and unchangeable by the routine — the five Talk:Family A–Talk:Family E pages of #58, listed for the sixth run, and the 37 superseded Proto-Andusian entries of #70, for the second. Both are deferred with the operator's word that the deletions will come; only an administrator can clear either.

Site totals: pages 1,264, articles 981, edits 3,807, files 1, job queue empty.

On the routine's own tooling. Every fetch was paginated through Python-built ASCII query strings with loop detection: the page fetch returned 1,264 of 1,264 titles, matching Special:Statistics, and the corpus's highest revision id was the last edit of the window, so nothing was measured against a stale copy. Two traps were re-hit and worked around rather than rediscovered: cargoquery returns etym_type as the key "etym type", with a space, so an underscore lookup silently reports zero — the finding of #72 depends on getting that right; and list=linterrors omits the query key entirely when there are no errors, which reads as a crash rather than as zero.

Self-check on this report. Its links were parsed before posting, reading the value of exists rather than testing for the key. The file is linked as [[:File:…]], never embedded. The four Tacchoan branch names and the island of deferred #57 are written unlinked here, as are the wanted Dictionary entries and the 37 titles queued for deletion, so this page adds no incoming link to anything it reports as wanted or as queued.

Pending

Seven rows, none high, all filed today: #72 #73 #74 #75 #76 #77 #78.

One is worth doing before the Dictionary grows again. #72 is a single mechanical pass over 270 pages, and it gets no cheaper: the three Andusian branch protos are registered and two of them have no entries yet, so every language seeded before this is settled adds another vocabulary whose descent the table cannot see. It is the same shape as the rows that sat pending through five runs while the corpus they described quadrupled.

Three close on one word each: #75 (link the twelve in the Members tables, or unlink them on the roster), #76 (is human in-frame vocabulary?) and #78 (does a file description restate a caption or point at it?). #74 needs a yes or no on whether polity means something country does not. #73 and #77 each need a name — a namespace for a test page, a category for the appendix pages.