Shelf

What's on it

Size

License

Classical Greek

Perseus: Homer, the tragedians, Herodotus, Plato, Galen… + 650 aligned English translations

1,418 docs / 394,706 passages

CC BY-SA

Post-classical Greek

First1KGreek: Athenaeus, Philo, church fathers, Swete's Septuagint

1,129 / 256,480

CC BY-SA

Classical Latin

Perseus: Vergil, Ovid, Cicero, Livy, Tacitus… + 181 English translations

534 / 391,799

CC BY-SA

Documentary papyri

Papyri.info DDbDP: contracts, letters, tax receipts from a millennium of Egypt (Greek, Coptic, Latin, Arabic)

61,414 / 921,611

CC BY

Latin inscriptions

Epigraphic Database Heidelberg: epitaphs, dedications, milestones from the whole empire — 81,416 dated (2026-07-14 census), genre/province/material facets

81,881 / 406,306

CC BY-SA

Coptic

Coptic Scriptorium: the complete Sahidic + Bohairic NT, monastic and patristic prose — gold-lemmatized (233k rows)

482 / 74,169

CC BY per doc (source class nc)

Sanskrit

GRETIL: Rāmāyaṇa, purāṇas, kāvya, śāstra, Ṛgveda with Vedic accents

780 / 703,068

CC BY-NC-SA

Treebanks

PROIEL, TOROT, UD, ISWOC: gold lemma/morphology/syntax — parallel NT ×5, OCS→Middle Russian, Old English, Old Irish glosses

80 / 178,278

mostly CC BY-NC-SA

Cuneiform

ORACC ×33 projects: the complete State Archives of Assyria, royal inscriptions, lexical lists, proto-cuneiform — with 8,911 aligned English translations

21,692 / 385,243

CC0

Biblical editions

Clementine Vulgate (73 books), SBL Greek NT, WEB English

184 / 81,372

PD / CC BY

Old English poetry

The complete ASPR: Beowulf, the Exeter Book, Dream of the Rood…

349 / 30,550

CC BY-SA

Slavic & Slovenian

CCMH OCS gospel codices, the ~1000 CE Freising Manuscripts, goo300k + IMP Early Modern Slovenian (1584–1899), the damaskini Balkan Slavic witnesses (15th–19th c., with English siblings)

839 / 456,189

CC BY (Freising BY-ND)

Celtic

CorPH Early Irish (gold-lemmatized: Annals of Ulster, the great gloss corpora), RIIG Gaulish inscriptions (with French siblings), the Ogham in 3D stones (real Ogham codepoints + transliteration layers)

1,387 / 20,318

CC BY / MIT (ogham nc pending clarification)

Chinese library

Kanripo classical Chinese + the CBETA Buddhist canon (Taishō + Xuzangjing) — 13.2M lzh passages (2026-07-22; the language now totals 18.7M with the Korean hanmun shelves), trad/simp/z-variant spellings folded to one search skeleton

13.2M passages

mostly CC BY-NC / CC BY

Japanese reading desk

Aozora Bunko: the Japanese public-domain library, Meiji core with a classical tail — ruby readings as annotations, kyūjitai reachable via the reform fold (2026-07-22)

17,195 / 2,991,807

open (PD grant)

Germanic wave

Menotec Old Norwegian treebanks + Poetic Edda, the Old Saxon Heliand (HeliPaD), ReM Middle High German (355k gold manuscript lines), Rundata runic inscriptions in five text lanes (2026-07-22)

31,057 / 409,947

nc / CC BY / CC BY-SA / odbl

Southeast Asia desk

The DHARMA corpora (Old Khmer, Campā, Nusantara, Pyu — with the first digital Old Mon anywhere), the Bagan Old Burmese inscriptions, the kakawin critical editions, the Old Javanese Wordnet (2026-09-02)

2,985 docs / 130,761 passages + 3,632 entries

CC BY / BY-SA

Korean desk

The Veritable Records of Joseon (sillok) in the original hanmun, the Goryeosa family, and since 2026-09-01 the OKHC historical slice: the Samguk sagi (1145), the Ilseongnok, the ITKC munjip mass, the AKS/Kyujanggak/NHM old-literature dbs

1,198,779 / 3,542,662 (okhc alone)

per-record PD/KOGL/nc (source CC BY-NC 4.0)

Reference shelf

LSJ + Lewis & Short + Bosworth-Toller + Monier-Williams + Wiktionary OCS + ten Wiktionary reconstruction/Celtic shelves + the IE-CoR / LIV / de Vaan etymological witnesses + the five StarLing bases (Pokorny, PIET, Vasmer, Germanic, Baltic) + three Slovenian historical dictionaries incl. Pleteršnik + the Hebrew/Egyptian/Slovene desks and the Sino-Japanese lexicography (Unihan, KANJIDIC2/JMdict, HDIC, Guangyun) (nabu define / etym)

1,310,763 entries / 56 shelves (2026-07-22)

CC BY-SA / CC BY / CC BY-NC-SA / grant

nabu search QUERY

FTS5 full-text search, bm25-ranked, diacritic-insensitive with per-language folding: μηνιν finds μῆνιν, iuvenis/juvenis/iuuenis all resolve. Filters: --lang, --license, --source SLUG (one shelf) and --axis NAME[,NAME…] (a research desk's shelves — the multi-source generalization of --source, e.g. --axis celtic), both composing with every other filter, --lemma/--near/--fuzzy included, the footer naming the desk; --limit. Date/place axis (163,821 dated/placed documents live — EDH inscriptions 81,416, HGV papyri, ORACC catalogue/regnal dates 21,558, TOROT chronicle annals, Slovene goo300k/IMP, Coptic manuscript dates — so --century -7 reaches the Assyrian letters): --from -300 --to -30 scopes by signed historical year (negative = BCE, no year 0), --century 6 is one century's shorthand, --place oxyrhynch% filters provenance — στρατηγ* --from 101 --to 300 --place oxyrhynch% finds the Oxyrhynchite strategoi. Genre facets (256,518 rows live from EDH): --type epitaph --province Britannia --material marble compose with a text query and all of the above. --kind CLASS[/SUB] filters by the ruled cross-corpus document-kind classes (1.7M classified documents): --kind funerary finds epitaphs across EDH, EDR, IIP and the rest regardless of each collection's own vocabulary (a head matches its whole family; --kind legal/contract narrows; --kind unmapped/unknown are first-class). Character-structure filters for the Han corpus (kept distinct from text FTS): --radical N (KangXi radical, Unihan), --strokes A-B (total-stroke range), --char-component C (characters containing C — KRADFILE ∪ BabelStone IDS transitive containment) AND together and compose with a text query as character-level filters on Han passages, the footer naming them. --words (Tibetan): keep only hits whose matched span aligns with WORD boundaries per the published segmentation model — an accidental syllable run across a word boundary no longer matches.

nabu search --lemma FORM

Dictionary-form search over 21.9M gold lemma rows in 45 languages (2026-08-31 census) — inflections, suppletion and all; hits carry glosses where the reference shelf knows the lemma. Add --morph case=dat,number=pl (UD feature vocabulary) to keep only attestations with that morphology, decoded evidence shown per hit — one façade over UD feats and PROIEL positional tags. Beside gold and the labeled silver shelves, the equivalence tier: CEIPoM's scholar-curated Classical-Latin keys on pre-Roman Italy's Oscan/Umbrian/Faliscan passages — search --lemma precor reaches the Iguvine Tables' pesnimu, every hit tagged [equivalence], never counted as attestation, --gold-only excludes.

nabu search --loans CODE

The language-contact facet (Coptic Scriptorium today): keep only passages carrying loanwords from a donor language — search ⲛⲟⲩⲧⲉ --lang cop --loans grc (131K+ Greek loan tokens tagged), composing with text, --lemma, --fuzzy, --near and every catalog filter. nabu list coptic-scriptorium --loans prints the donor-language census; --loans grc enumerates the most loan-saturated documents.

nabu search A --near B [--window N]

Proximity search: keep only hits where B is within N words of A in the same passage (FTS5 NEAR over the folded forms; default 10, 0 = adjacent, order-independent). λόγος --near θεός is John 1:1; composes with --lemma (the anchor expands to the lemma's attested surface forms first: --lemma λέγω --near κύριος finds τάδε λέγει κύριος) and --lang/--license/--limit. Both terms bracketed in the snippet.

nabu search --fuzzy FRAGMENT

Damaged-text fragment search: substring matching ANYWHERE in a passage, mid-word included — ']μηνιν αει[' works typed straight off the edition (editorial brackets stripped, then the same per-language folding as plain search). Character-trigram index over the DOCUMENTARY shelves only (papyri-ddbdp + oracc + edh, fuzzy_index: true in the registry — corpus-wide would cost 15×), candidates verified by real substring match; every render names the indexed scope. The production index is LIVE (1,713,135 passages indexed as of 2026-07-14, EDH aboard). Fragments need ≥3 characters; composes with --lang/--license/--limit/date-place filters; --long prints the whole folded passage. For literary half-memories use plain search or parallels.

nabu show URN

A passage, a whole document, or a citation range (urn:…:1.1-1.10) with license, revision, and full provenance trail. --parallel pairs the aligned English translation; --random pulls something off the shelf; --segmented renders Tibetan text word-split (tsheg delimits syllables, not words — the boundaries come from Nabu's own published segmentation model, nabu sync nabu-data).

nabu parallels URN

Passage-anchored intertext: point at one passage and find where the corpus quotes or echoes it — reception discovery, not translation alignment. Query-time over the FTS index (no new schema): the anchor's 4-word grams are phrase-probed, candidates ranked by shared-gram count weighted by rarity, elision folded across editions (so Matthew 4:4 finds LXX Deuteronomy 8:3). One hit per document (duplicate witnesses grouped, loci counted), the shared phrase shown as evidence; a gold-lemmatized anchor also gets rare-lemma "echoes" (re-inflected allusion). --long expands the truncated evidence; --lang/--license/--limit scope. --batch SCOPE flips the engine to corpus-wide mining: every anchor of a source slug or urn prefix, hits persisted as kind=parallel edges in the links journal (top --per-anchor per anchor at --min-score+, both named in the summary — no silent caps); reruns supersede, interactive output never persists.

nabu formulas SCOPE

The oral-formulaic reader's mirror of parallels, pointed inward: mine the repeated formulas WITHIN a corpus slice (a source slug or a work/urn prefix) — the same gram machinery, counting instead of probing. Homer's ὣς ἔφαθ' οἵ δ' (72×), τὸν δ' ἀπαμειβόμενος προσέφη πολύμητις Ὀδυσσεύς (50×); the Old English saga hwæt ic hatte riddle refrain, Beowulf maþelode bearn Ecgþeowes, awa to feore. Ranked by count × length — no stoplist, the ranking is self-filtering (a genuine formula out-recurs any function-word run; measured). --lang mines one tradition where a source mixes translations; --gram-size, --min-count, --limit, --long (every locus). Zero schema, ~0.2 s per slice. --batch SCOPE persists the sweep as kind=formula edges: a STAR per formula (hub = its first locus in urn order, one edge to every other locus, the gram riding each edge's detail, the count its score — all-pairs would explode quadratically), top --max-formulas by rank; reruns supersede, interactive output never persists.

nabu links URN

The mined cross-reference graph, read back: every batch-produced edge touching a urn, both directions, grouped by kind (parallel, formula, cognate), counterparts resolved to title/language, each kind's evidence rendered natively (a parallel's score, a formula's “gram” ×count pointing at the refrain's hub, a cognate's ref · root [shelf] meet), provenance footer citing the producer run (scope, params, code version, date). Edges are urn-keyed in their own journal (db/links.sqlite3) so they survive nabu rebuild untouched; show grows a one-line linked: N formula, M parallel footer counting the kinds present when edges exist. --long lists every edge per kind.

nabu align REF

One citation across every witness of a registered work (config/alignments.yml) — the parallel NT and the Septuagint ↔ Vulgate OT ship as flagships. A whole-chapter or verse-range query clips at 200 refs by default; --long lifts that ceiling and renders every ref. --collate diffs the witnesses into a compact apparatus (base reading + per-witness divergences) per (language, script) group, with cross-script witnesses rendered undiffed and labelled honestly; --base LABEL picks the base.

nabu define LEMMA

Dictionary-shelf lookup — LSJ, Lewis & Short, Bosworth-Toller, Monier-Williams (193,890 entries live: SLP1 transcoded to IAST, so define amsa reaches aṃśa/aṃsa, RV./BhP. citations resolving into the GRETIL shelf at verse grain, MW's own Gk./Lat./Goth. cognate notes feeding etym), Wiktionary OCS — entry citations resolved to in-catalog passages/documents. The TLS Chinese shelf resolves at SENSE grain: define 棄 lists where each sense is attested in the classics (189K sense-level attributions), the Kanripo-held texts resolving to page or document urns and coverage growing as the kanseki waves sync. A leading * scopes to the seven reconstruction shelves (Proto-Slavic/PIE/Proto-Germanic plus the four intermediates), whose entries list their descendant reflexes; --long expands the truncated "not attested here" list in full, grouped by language. Query expansion reads Nabu's own published datasets back: a Sanskrit surface form reaches its DCS gold lemmas, and a Tibetan verb stem reaches its paradigm lemma (ཕྱིན or phyin finds འགྲོ — suppletion only the published verb table knows).

nabu etym LEMMA

The comparativist's walk: an attested lemma (богъ, guþ) → every reconstruction whose Wiktionary descendants name it → one hop up the proto-to-proto chain, each with cognates and corpus attestation counts. Multi-hop closure (PIE *per- → PBS → *pьrstъ → chu/orv in one walk) and per-edge "(loan)" flags are live, and so — since their 2026-07-14 first syncs — are the three independent expert-curated witnesses: IE-CoR's cognate sets ride the same walk (etym срьдьцє → *k̑erd- with grc καρδία ~ lat cor ~ got hairto beside kaikki's *ḱérd-, curated loan events labelling their edges (loan)), LIV supplies the PIE verbal roots, and de Vaan's EDL the lat → Proto-Italic → PIE Leiden chains. --long expands every truncated cognate list, grouped by language with each code named inline ([gkm · Medieval Greek]; compact is the default).

nabu language CODE

The code desk reference: any language code the library surfaces — corpus tags (chu, san-Latn) and the 803 Wiktionary etymology codes in etym's cognate lists (gkm, zle-ort, zlw-opl) — explained on one card: name (from the derived names census, filled at the owner's 2026-07-14 wiktionary resyncs), family, curated historical context (file-backed since P19-1: one Markdown dossier per code under local/shelves/local-language/, 199 live — edit in any editor, nabu sync local-language re-derives the card), and live holdings (documents/passages, gold-lemma rows, dictionary shelves, etymology edges; zero fields suppressed). An unknown code misses honestly with a family hint. --list prints the held languages plus the nabu-lects registry extension (every anchor's minted stage/variety ids — grc:koi, ave:old — the vocabulary --lect and the postures accept); --long adds per-source counts and the upstream-code edge split.

nabu info QUERY

The dossier front door: one command answering "what is this?" across the three desk namespaces, dispatching to the existing cards byte-identically — a single character or Gardiner-style sign code → the char card; a known language code or lect id → the language card; a script tag or name (glag / Glagolitic) → the curated script dossier (the same hand-written context published as nabu-data's mul/script-dossiers), with the registry's scripts-table row beneath. Earlier rungs win, shadowed readings are footnoted, and anything else misses honestly naming all three namespaces.

nabu axis [NAME]

The research-axis desk reference (config/axes.yml) in the nabu language mold: bare, every desk with its persona one-liner; nabu axis celtic, one desk's full card — persona verbatim, membership rationale, every member source with its enablement and live holdings (documents/passages, entries, languages, license mix; a member holding nothing says so), the aggregate gold-lemma coverage across the desk's held languages, and the shipped affordances (list --axis NAME, sync NAME). An unknown axis misses naming the known set.

nabu char CHAR

The character desk card: one Han character composed from every shelf the library holds — the join no dictionary site offers. Header (glyph · total strokes · KangXi radical number+name, from Unihan) → IDS decomposition with each component's follow-up commands (BabelStone IDS) → the component index (KRADFILE) → variants (trad/simp/z-variant, Unihan) → readings (kun/on + meanings from KANJIDIC2; Mandarin/Korean/Vietnamese from Unihan) → pedagogy (Jōyō grade · JLPT · frequency) → desk-reference codes (Unicode always; four-corner/SKIP/JIS/dic where KANJIDIC2 holds them) → the diachronic column no online dictionary carries (Baxter-Sagart Old Chinese · Qieyun Middle Chinese · the Heian hanzi dictionaries · TLS senses with classical attestation counts) → corpus attestation → the search CHAR* follow-ups. Matches Jisho field-for-field where a shelf backs it, exceeds it diachronically, and never renders an unbacked field (absent, not "—").

nabu signs URN|TEXT

The cuneiform sign desk: ATF transliteration — a passage/document urn from the ATF shelves (cdli/oracc/ebl/etcsl) or raw pasted text — resolved token by token through the local Oracc Sign List (nabu sync osl) into sign identities: value → sign name → Unicode codepoint(s)/glyph. C-ATF ASCII folds (sz→š, s,→ṣ, u2→u₂) and the ETCSL romanization (c→š, j→ŋ; auto for etcsl urns, --dialect etcsl for raw text). One honest status per token — deterministic · qualified · ambiguous (all candidates listed) · no-codepoint (sign real, honestly unencoded) · broken · unknown (not in the OSL, said plainly); determinatives unbraced and marked; absent data absent, never "—". --json emits the frozen machine contract the MCP nabu_signs tool serves identically — the downstream boundary: the sign list adds identity, not curriculum.

nabu cognates TARGET

Cognates in parallel: verses of an alignment work where witnesses in ≥2 languages use reflexes of the same reconstruction root — the hub × crosswalk join (cognates nt --langs got,chu → saltсоль at *sḗh₂l, hlaifsхлѣбъ at *hlaibaz, ~300 verses / 30 roots in under a second). TARGET is a work id, verse, chapter, or book; every hit names its meet SHELF (a gem-pro meet for a Slavic word reads as a borrowing); corpus-common words suppressed with an honest count (--all lifts); --long lifts the 200-hit cap and expands detail. --batch WORK persists the whole-work map as kind=cognate edges between the cross-language witness passages of each meet, the meet itself (ref · root [shelf]) riding each edge's detail; reruns supersede, interactive output never persists.

nabu concord QUERY

Classic KWIC concordance: keyword column-aligned in pristine text, corpus order — for scanning usage, not relevance.

nabu kind census

The document-kind axis (the fourth axis beside dates, places, and lects): what KIND of document is each text — funerary, administrative, legal, letter, divination, historiography… Thirteen collections' own genre vocabularies (sepulcralis / epitaph / funerary / Inscription funéraire) fold onto ~26 ruled cross-corpus classes, multi-label (an ode to a ruler is poetry AND royal), every upstream label preserved verbatim. census shows the class board with document/source spread, the honesty buckets (unknown = upstream's own "cannot determine"; unmapped = values awaiting a fold rule — --unmapped lists them per source), and the unclassified remainder. Filter with search --kind; a document's classes render on its show card.

nabu vocab URN

Lemma-frequency profile of a document, range, or passage against the gold-lemma corpus: total tokens, distinct lemmas, the most distinctive vocabulary (log-odds vs corpus — Caesar surfaces legio/proelium, Cicero's De officiis surfaces officium/honestas), and the in-document hapax legomena. Gold shelves only; a document without gold lemmas says so and names the annotated languages. --long lists every hapax (and every gold-bearing language) in full, escaping the --limit display cap. --by-century switches to diachronic mode: the shape of the dated corpus over time, or — with a text query — a word plotted across the centuries (vocab --by-century 'στρατηγ*' --lang grc peaks in the 2nd c. CE), bucketed by earliest year and honest about ranges that span more than one.

nabu lemma FORM [--lang CODE]

A lemma's attestation profile from the precompiled lemma rollup: first and last attested century, each with its date basis (artifact vs composition dating), a per-century table and sparkline with gold (verified annotation) and silver (automatic lemmatization) kept apart, undated attestations listed last, the top sources, and the dictionary shelves carrying the form. A document counts at the midpoint of its date envelope. Companions on the language desk: nabu language CODE --curve (documents + passages per century) and nabu language --mixing (the multilingualism census — documents by passage-language inventory size, the commonest mixed inventories, and per source whether mixing runs in blocks or interleaves). All three read only precompiled rollups, rebuilt with the library and by rake builders:refresh.

nabu export --format plain|jsonl

Stream the corpus out, with --lang/--license/--source/--axis NAME[,NAME…] filters (--axis is the multi-source generalization of --source — a whole research desk's shelves) — the longevity-hedge exit formats.

nabu ingest FILE-or-URL...

The intake front door for your own material: copies a PDF/scan/article into the local-library shelf (research_private by default — never served or redistributed) — or downloads an http(s) URL first, recording it in the manifest — derives metadata candidates mechanically, confirms them interactively / AI-assisted (--assist CMD) / scripted (--yes), then syncs and prints the minted urn. --shelf language CODE scaffolds a language dossier, --shelf source SLUG a source dossier.

nabu note URN [TEXT]

Owner annotations — scholia of one's own — on any urn the corpus knows (documents, passages, ranges, dictionary entries), resolution-checked before any write, stored as plain YAML on the local-notes shelf. Bare nabu note URN reads back what you said; --list enumerates; --force records a deliberately dangling note on planned material. Notes render on show/define/links and are served over MCP with their target's withholding rules.

nabu enable AXIS|SOURCE...

The first-time step before sync: adds sources and/or whole desks to local/config/profile.yml, this box's enablement config — enablement governs what sync/sync --all acquires and the default row set of list/status/health (query commands always see everything held). Grant-gated sources prompt for the recorded acknowledgment (--grant-acknowledged for scripted use). Browse the menu first with status --disabled / list --disabled — the complement view: only the rows enable would add (--all is everything).

nabu sync SLUG / sync --all

Fetch and load a source (git, zip, or single-file HTTP — or re-scan a local shelf); idempotent, non-destructive, every run recorded. The name resolves slug-first-then-axis: sync celtic (or --axis a,b) expands a research axis to its enabled members, grouped, and names any unwired members on one skip line — feature modules on their own skipped (module — sync directly) line — (an axis is not an explicit request, so an unwired member is skipped where an explicit sync SLUG would sync it anyway).

nabu list [SOURCE]

The what-is-held view (status is the sync-state view): bare, a content census — one line per shelf with document/passage/entry counts, languages, the effective license-class mix, withdrawn/retired counts when nonzero. With a SOURCE, one shelf's card (identity, credit line, counts, per-language breakdown, dictionaries, date-axis coverage, facet and collection summaries). --documents / --entries / --collections enumerate (default --limit 50, 0 = all, honest "… N more" tail), with --lang/--license/--withdrawn/--from/--to/--century filters on documents and --prefix folded headword-prefix filtering on entries (bh finds *bʰer-). --axis groups the census under the research axes (the owner's desks) — bare, every axis in ratified order with its persona line; --axis slavic or --axis a,b selects some — a source appearing under each axis it serves.

nabu status / health / verify

Per-source counts and run history, each row carrying an up= upstream-drift column (up=ok(2d) / up=BEHIND(2d) / up=stale(30d) / up=?(never) / up=?(re-probe) when a cached verdict predates the last sync / up=frozen) so an update is an informed decision — nabu status --remote probes upstreams inline and refreshes it in one command; local trend + upstream drift checks; full bitrot/tamper re-verification of every canonical file. health also runs the mechanical postcondition invariants (P18-7): failed-run/partial-load surfacing, flag-vs-artifact and synced-vs-populated mismatches, pending migrations, and quarantine counts as a DELTA against an audited baseline — plus an optional sync --review CMD AI-review hook, off by default. Its default view prints findings only (ok rows and by-design notes fold into one rollup; nabu health --all shows every row).

nabu mcp

The read-only MCP server — fourteen tools for Claude Code/Desktop and any MCP client. Recipes and the user-perspective walkthroughs in docs/mcp.md.

Doc

One line

docs/quickstart.md

Zero to first search, copy-pasteable, honest about sizes and timings.

docs/library.md

The shelf map: every corpus with contents, counts, licenses, and research uses.

docs/axes.md

the research axes: the twenty-four scholarly desks, their personas, and which sources each tags.

docs/01-concept.md

The vision: what Nabu is, workflows, principles, what success looks like.

docs/mcp.md

The MCP server: twelve read-only tools, registration recipes, quoting etiquette.

docs/conventions.md

Field notes for working with ancient-text corpora (Unicode/NFC, citations, editions, licensing) — start here if you're new to the domain.

docs/architecture.md

The design: layer model, adapter contract, store schema, retention machinery.

docs/02-sources.md

The source inventory: every corpus scouted, scored, and license-checked.

docs/03-unlockable-sources.md

Sources not ingestible today, with concrete unlock paths.

docs/ops.md

The runbook: maintenance cadence, launchd templates, what to do when a check goes red.

docs/maintenance-and-extension.md

How this stays alive across years of intermittent attention — leads with the where-truth-lives map: which surface is authoritative for every aspect, and what merely projects it.