Examples

Worked walk-throughs, one per discipline, using real commands and output from live runs between July and September 2026 (each section notes its date where it matters). Nothing below is a mock-up; trims are marked with ellipses. Each discipline is also a research desk — one of the 25 research axes, where that desk’s shelves, instruments, CLI recipes and terminal setup live on one page.

For the classicist

The Classical desk.

The Perseus Greek and Latin canons and First1KGreek put 2,209 Greek and Latin editions on the desk, 872 of them with aligned English translations — show <urn> --parallel pairs Vergil or Homer line by line (an example appears on the Tools page). Proximity search supports collocation questions in the manner of the TLG interface, and is aware of lemmas, so suppletive paradigms do not escape it. A simple example — where does λόγος stand within five words of θεός?

$ bin/nabu search λόγος --near θεός --window 5 --lang grc
urn:nabu:ddbdp:p.oxy:8:1151:18   [θεοσ] ην ο [λογοσ].          ← a papyrus amulet…
urn:cts:…:tlg0031.tlg004…:1.1    …και [θεοσ] ην ο [λογοσ].     ← …quoting John 1:1

The first hit is not a literary text at all: it is a Christian amulet from Oxyrhynchus quoting the opening of John — the kind of cross-shelf find (papyri beside literature in a single index) that motivated the library’s design.

For the Latinist, across a millennium and three tiers

The Classical and Romance desks — live run of 3 September 2026.

Latin in this library runs from archaic dedications to Migne’s Patrologia, and lemma search answers across all of it — with every annotation honestly labelled by who made it. One dictionary form, three tiers in three hits:

$ bin/nabu search --lemma gratia --lang lat --limit 3
urn:nabu:ceipom:88:97 [lat] [equivalence]  gratia → gratia  (Favor which one finds with others)
  Orcevia Numeri [uxor] / nationu(s) gratia / Fortuna Diovo fileia / Primogenia / donom dedi
urn:nabu:corph:0016:S0016-2 [lat]  gratia → gratias  (Favor which one finds with others)
  Finit amen deo gratias ago oroit do dimmu . , . .
urn:nabu:digiliblt:dlt000003:116 [lat] [silver]  gratia → gratia  (Favor which one finds with others)
  gratia et bonae et malae rei causa;
3 hits (exact lemma match; text is pristine) — 1 silver (automatic lemmatization; --gold-only excludes) — 1 equivalence (scholar-curated Classical-Latin equivalents; --gold-only excludes)

An archaic Praeneste dedication answering through a scholar-curated [equivalence], an Insular colophon on gold annotation, a late-antique grammarian through machine ([silver]) lemmatization — and the summary line tells you exactly which is which. --gold-only restricts to the human-verified tier; the tier system itself (21.9M gold, 35.9M labelled silver annotations) is written up in lemma enrichment.

Historical stages are a filter of their own — --lect lat:med keeps only shelves resolving to Medieval Latin, so a formula query stays inside its period:

$ bin/nabu search misericordia --lect lat:med --limit 2
urn:nabu:corpus-corporum:8706:d15.p1 [lat]
  …mino quoniam bonus: quoniam in aeternum [misericordia] ejus. Confitemini Deo deorum: quoniam i…
urn:nabu:corpus-corporum:9349:d142.p1 [lat]
  …mino quoniam bonus, quoniam in aeternum [misericordia] ejus. Confitemini Deo deorum, quoniam i…

For the papyrologist

The Epigraphy desk — papyri, stones and sherds.

The library holds 61,414 documents of the Duke Databank of Documentary Papyri, with the Leiden editorial conventions preserved. Fragment search is built for damaged texts: type the line as the edition prints it, brackets and all, and the trigram index matches it inside words:

$ bin/nabu search --fuzzy ']ανδρα μοι εν['
urn:nabu:ddbdp:bgu:6:1470:ctr:6 [grc]
  μαρτυροι. [ανδρα μοι εν]νεπε μουσα πολυτρο
1 hit (fuzzy substring; highlights are diacritic-folded)
fuzzy index covers: oracc, papyri-ddbdp

— BGU 6.1470, a Hellenistic writing exercise that breaks off mid-word through the Odyssey’s opening line (…Μοῦσα πολύτρο[πον). Since the papyri carry HGV dating and provenance, chronological and geographic filters compose with any search: search 'στρατηγ*' --from 101 --to 300 --place oxyrhynch% scopes to the Oxyrhynchite strategoi of the second and third centuries.

For the sinologist

The Sinitic desk — live run of 3 September 2026.

Classical Chinese writes without word boundaries, which defeats ordinary full-text engines; since September 2026 the library maintains a dedicated character-pair index over the Sinitic, Korean and Japanese shelves, so a Han query is exact, fast, and never depends on guessed segmentation. The six dragons of the Yijing’s first hexagram, across three corpora in one query:

$ bin/nabu search 六龍 --limit 3
urn:nabu:classical-modern:汉书:志:礼乐志:298 [lzh]
  吾知所乐，独乐[六龙]，六龙之调，使我心若。
urn:nabu:cbeta:X37n0675:0802a05 [lzh]
  槃如晝夕窹，遠離夢想也。倒正如[六龍]舞，以六龍
urn:nabu:cbeta:X86n1600:0170a04 [lzh]
  乘[六龍]以御天。知時之乘六龍，則知一句之具三
3 hits (snippet shows the text as stored; matching is fold-aware)

— the Book of Han’s music treatise (in simplified-character transmission, found by the same fold-aware query) beside two Buddhist commentaries quoting the Yijing line 乘六龍以御天, “riding the six dragons to drive across the sky.” The character itself then goes on the desk — one card from modern readings back to Old Chinese:

$ bin/nabu char 龍
龍  U+9F8D  ·  16 strokes  ·  radical 212 龍 dragon
…
readings (ja, KANJIDIC2):
  on: リュウ、リョウ、ロウ
  kun: たつ、いせ
…
readings (sinoxenic, Unihan):
  Mandarin: lóng
  Korean (hangul): 룡:0E 용:0
  Vietnamese: long
…
Old Chinese (Baxter-Sagart):
  [baxter-sagart-oc] dragon
    OC: *[mə]-roŋ
    MC: ljowng (l- + -jowng A)
    pinyin: lóng
…

For the slavist

The Slavic desk.

The Old Church Slavonic canon is held complete across its editions: Marianus, Zographensis, Assemanianus, Savvina kniga, and Suprasliensis (folio-line cited, with hyphen-split words searchable whole), beside Old East Slavic from birchbark letters to Ruthenian chancery texts, and the ~1000 CE Freising Manuscripts in three aligned transcription layers. The gospel codices join the New Testament alignment hub, so align REF --collate turns the witnesses into a working apparatus — a raw-token diff per script family, with the four Helsinki-transliteration codices collated against each other and the Cyrillic witnesses set beside them, honestly uncollated, because a mechanical fold cannot bridge the two transcription systems. One verse across the tradition:

$ bin/nabu align "MARK 2.3"
…
marianus — Codex Marianus [chu]   license: nc
  urn:nabu:proiel:marianus:36421
    Ꙇ придѫ къ немоу носѧште ослабленъ жилами. носимъ четꙑрьми.
…

For the comparativist

The Etymology desk.

The reconstruction shelf walks attested words to their proto-forms and cognates, with corpus attestation counts at every step:

$ bin/nabu etym богъ --lang chu
богъ [chu] → *bogъ [sla-pro] — gloss: god
← *bʰeh₂g- [ine-pro] — gloss: to divide, distribute, allot
  reflexes: [grc] ἔφᾰγον, [sa] भक्ष (bhakṣá), …

Pure-ASCII input works for reconstructed forms (etym bhewgh). The cognates command then crosses this crosswalk with the alignment hub: verses where the witnesses use reflexes of the same root, found without any surface resemblance —

$ bin/nabu cognates "LUKE 14.34" --langs got,chu
LUKE 14.34  *sḗh₂l [ine-pro · attribution]
    chu  соль — attested as солъ
    got  salt

The whole Gothic × OCS New Testament yields roughly 300 such verses across 30 roots in under a second (hlaifs ~ хлѣбъ, malan ~ млѣти, menoþs ~ мѣсѧць), and each hit is labelled with the dictionary shelf on which the two languages meet — a Proto-Germanic meet for a Slavic word reads as a likely borrowing, not common descent.

For the Old English scholar

The Germanic desk — live run of 3 September 2026.

The Anglo-Saxon Poetic Records sit beside the prose, the Bosworth-Toller dictionary, and the runic and continental Germanic shelves. The formula miner reads oral-formulaic poetry the way its scholarship does — by finding the repeated building blocks and counting them:

$ bin/nabu formulas aspr --min-count 5
formulas in aspr — 30550 passages / 175736 tokens
13×  ic waes ond mid
     e.g. urn:nabu:aspr:A3.11:59, urn:nabu:aspr:A3.11:60, urn:nabu:aspr:A3.11:61
12×  ond thaet word acwaeth
     e.g. urn:nabu:aspr:A2.6:1071, urn:nabu:aspr:A3.1:316, urn:nabu:aspr:A3.1:474
12×  saga hwaet ic hatte
     e.g. urn:nabu:aspr:A3.22.3:72, urn:nabu:aspr:A3.22.8:8, urn:nabu:aspr:A3.22.10:11
…

Saga hwæt ic hatte — “say what I am called” — is the Exeter Book riddles’ closing refrain, surfacing here with the exact riddles that use it; the speech formulas (ond þæt word acwæð) rank right beside it. The Germanic ladder is also growing downward in time: the Deutsches Textarchiv adapter (the German print corpus, 1473–1969) is built and awaiting its first sync.

For the assyriologist

The Cuneiform desk.

The cuneiform shelf now holds 104,722 ORACC documents (CC0) across 38 projects — the complete State Archives of Assyria, the ePSD2 corpora with the Ur III administrative mass, the Achaemenid trilinguals — with upstream gold lemmatization feeding search --lemma for Akkadian and Sumerian. Beside it sit the CDLI’s 353,156 catalogued artifacts (135,201 of them transliterated: proto-cuneiform and proto-Elamite’s only machine-readable home), the eBL Fragmentarium’s 23,288 fragments with inline English, and the Sumerian literary canon in two scholarly editions (ETCSL beside ePSD2, deliberately unmerged, meeting at reference edges — all synchronized 19 July 2026). A state letter reads with its facing English, line-anchored:

$ bin/nabu show urn:nabu:oracc:saao-saa01:P224395:o.1-o.3 --parallel
:o.1  akk  a-na LUGAL EN-ia
:o.2  akk  ARAD-ka {1}10-ha-ti
…     eng  To the king, my lord: Your servant Adda-hati. …

And the shelf rewards browsing:

$ bin/nabu show --random --source oracc
urn:nabu:oracc:rinap-rinap1:Q003443:1 [akk]
  a-di {KUR}sa-u₂-e KUR-e ša ina {KUR}lab-na-na-ma it-tak-ki-pu-u₂-ni
  document: urn:nabu:oracc:rinap-rinap1:Q003443 — Tiglath-pileser III 30
  source: oracc   license: open   sequence: 0   revision: 1

— a royal inscription of Tiglath-pileser III, “as far as Mount Saue, which abuts Lebanon.”

For the hittitologist

The Hittite desk.

TLHdig (Hethitologie-Portal Mainz) brings >98% of the published Hittite tablet fragments: 23,486 manuscripts in 663 CTH compositions, each line carrying transliteration, Unicode cuneiform, and upstream’s candidate morphological analyses — served at an honest silver tier (only upstream-disambiguated analyses mint searchable lemmas; the rest ride as annotations). The UD HitTB treebank adds 1,309 gold-lemmatized words from Hoffner & Melchert’s grammar, and the Anatolian reflex columns of the comparative shelf light against the corpus:

$ bin/nabu search --lemma ḫūmant --lang hit     # "all/every" across the corpus
urn:nabu:tlhdig:101:tlh:kub.21.9:7 [hit] [silver]  ḫumant → ḫu-u-ma-an-da
  e-e]p-ta KUR.KUR°ḪI.A°-ia-ši ḫu-u-ma-an-da za-aḫ-ḫ[e-er]
…

$ bin/nabu etym water                           # the textbook heteroclite, now corpus-attested
  ← *wódr̥ [ine-pro] — Wiktionary — Proto-Indo-European …
  [hit] 𒉿𒀀𒋻 (wātar) — silver 704 passages

Both runs are pasted live (19 July 2026, trimmed lines marked).

For the biblical scholar

The Biblical desk — one text across every witness language.

The New Testament is held in up to fifteen registered witnesses — two Greek editions, two Latin, Gothic, Classical Armenian, five Old Church Slavonic manuscript editions, Old English, English, and (since 13 July 2026) Sahidic and Bohairic Coptic — and one command renders any verse across all of them, each row carrying its license label (the run below predates the two Coptic witnesses; the full output opens the Home page):

$ bin/nabu align "MARK 2.3"
MARK 2.3 — New Testament (parallel witnesses)
  13 of 13 witnesses attest this ref
…

The Old Testament axis grew its Hebrew legs on 18 July 2026: align "GEN 1.1" now renders the Masoretic text (byte-verbatim Leningrad Codex, twice — OSHB and the ETCBC’s BHSA) beside the Septuagint, the Vulgate, the English, and Targum Onkelos’s Aramaic — six witnesses per verse, with the Greek/Hebrew Psalm numbering divergence mapped rather than hidden: align "PSA 22.1" shows the Hebrew witnesses’ 23.1, labelled as such. Dictionary lookups close the loop — the LSJ entry for μῆνις resolves its citations to live passages in the catalog (see Tools).

For the egyptologist

The Egyptian desk.

Three millennia of Egyptian arrived on 18 July 2026 as one axis: the TLA/BBAW sentence corpus with gold lemmatization, its dictionary keyed by the same lemma ids, the only bulk demotic dataset in existence, and the Coptic lexicon with a crosswalk that walks a word across the whole span.

$ bin/nabu show urn:nabu:aes:sawlit:BRMYDZFU3BFT7JLX45UAGVMKMI --parallel ger
  :IBUBd5wLLFaLn06ZvzVW4ePuFIg
    egy  šms,w Zꜣ-nh,t ḏd =f
    ger  Der Gefolgsmann Sinuhe, er sagt:
…
$ bin/nabu search --lemma nfr          # every inflection, glossed
$ bin/nabu show urn:nabu:dict:aed:tla866216   # the root nfr and its 56 derivatives
$ bin/nabu links urn:nabu:dict:ccl:C1494      # ⲕⲁϩ ← qꜣḥ ← qh, 3,000 years

For the student of pre-Roman Italy

The Italic desk.

Seven fragmentary languages landed 18 July 2026, most in their only machine-readable form: the oldest Latin in existence, the complete Iguvine Tables in Umbrian, Etruscan with a scholarly glossary, Oscan in Greek script from Messina, and the Alpine corpora of Lepontic and Raetic.

$ bin/nabu show urn:nabu:ceipom:2:2
  Manios med fhefhaked Numasioi         # the Fibula Praenestina, 7th c. BCE
$ bin/nabu show urn:nabu:ceipom:995     # the Iguvine Tables, all 688 sentences
$ bin/nabu define avil                  # Etruscan "year" (ETP glossary)
$ bin/nabu etym pompeii
  Pompeii [lat] (loan) → *pompe [osc] ← *kʷenkʷe [itc-pro] "five"

For the Hebraist

The Hebrew desk — and its terminal setup for right-to-left reading.

The Masoretic text is held byte-verbatim — the Leningrad Codex’s combining-mark order is never normalized away — with full morphology, ketiv/qere both preserved, and the verse-aligned Aramaic of the Targums beside it. The ETCBC’s BHSA adds clause-level syntax, the Dead Sea Scrolls (Abegg’s transcriptions, 1,001 scrolls) sit beside the codices, and the augmented-Strong’s lexicon resolves every OSHB lemma to its BDB entry — all live since 19 July 2026.

$ bin/nabu show urn:nabu:oshb:gen:1.1
  בְּרֵאשִׁ֖ית בָּרָ֣א אֱלֹהִ֑ים …
$ bin/nabu align "GEN 1.1"              # seven witnesses: MT ×2, LXX, Vulgate, English, Onkelos, Peshitta

A note on scope: the library today serves the Hebrew Bible and Second Temple corpora. Rabbinic literature proper — Mishnah, Talmud, midrash — is a planned campaign of its own; the Targum shelf and the Jastrow dictionary scans are its first bridgeheads.

For the southeast asianist

The Southeast Asia desk — the newest desk (September 2026); live run of 3 September 2026.

The Indic cosmopolis at its eastern edge: the DHARMA project’s Old Khmer, Cham, Old Javanese, Pyu and Nusantara epigraphy, the Bagan inscriptions of Old Burmese, and the Old Javanese Wordnet as the desk’s glossary. The shelf rewards a blind draw — every record carries its edition credit and resolved language stage:

$ bin/nabu show --random --source dharma-khmer
urn:nabu:dharma-khmer:INSCIK00299-24:1 [okz-Latn]
  .kṣuradhāraparvvata . qnak· ta lvac· tamrya qseḥ yāna . pāduka . cap· ta vrāhmaṇa nu jeṅ(·) …
  document: urn:nabu:dharma-khmer:INSCIK00299-24 — K. 299-24. Eastern section of southern gallery of Angkor Wat (the hell and heaven gallery), 12th century
  source: dharma-khmer   license: attribution   sequence: 0   revision: 1
  credit: DHARMA — Corpus des inscriptions khmères (the Cœdès K-numbers) (ERC no. 809994, EFEO/CNRS; erc-dharma on GitHub). Cite the corpus and its editors wherever used.
  lect: okz~latn (codemap)
…

— the hell-and-heaven gallery of Angkor Wat, naming the razor-edged mountain Kṣuradhāraparvata and the sinners it awaits: elephant thieves, horse thieves, and those who show contempt for brahmins and pandits.

Searching by meaning

Built September 2026; the vector store’s first overnight build is scheduled.

The newest search layer answers a question no word index can: where else does the library say this? — same thought, different words. search --similar <urn> anchors on any passage of the literary core and returns its nearest same-language neighbors — another edition’s copy of the same line, a quotation drifted by memory, a paraphrase — ranked and banded close / near / loose, with the embedding model named on every page so a statistical match is never mistaken for a curated alignment. The same lane serves AI conversation through the MCP server’s similar_to parameter. It runs on a local vector store the owner builds once with nabu embed and tops up incrementally after every sync; no text ever leaves the machine. The plain-language write-up is embed.md — and this page will gain the live walk-through once the first build lands (nothing here is ever pasted from imagination).

Composing the layers

Added August 2026, live runs of 12 August — the place, script, sign, and dating layers answering one question together.

The library’s layers — places as identities, writing systems, sign readings, dates — long had desks of their own; since August 2026 they compose in search. Ask for a common word at one findspot, by identity rather than spelling, and get the same page every time (--scan walks the filtered documents in corpus order — deterministic, never sampled):

$ bin/nabu search lugal --place Girsu --scan --limit 5
urn:nabu:oracc:dcclt:P217451:f.1 [sux]
  {d}nisaba [lugal]-ušumgal dub-sar ensi₂ lagaš{ki}
urn:nabu:oracc:epsd2-admin-ur3:P100144:o.1 [sux]
  3(aš) 4(barig) še gur [lugal]
…
5 hits (… corpus-order scan of the filtered set (--scan))
note: place: "Girsu" → cigs:GIR = pleiades:912855 = cdli-provenience:88 = geonames:93976 (identity refs + name match)

The note is the point: “Girsu” resolved to the place’s identities across four gazetteers (the crosswalk the Places page describes), so tablets citing any of those refs — in any spelling — answer, alongside plain name matches. The place card shows the same equivalences:

$ bin/nabu place cigs:GIR
Girsu, cigs:GIR — 31.56, 46.18
  = cdli-provenience:88 = geonames:93976 = pleiades:912855
  axis holdings (place_ref — adapter-asserted + registry decisions): 19226 oracc

A cuneiform sign searches as itself — its reading values expand into one query, and each hit brackets the value it matched:

$ bin/nabu search --sign 𒊬 --lang sux --limit 3
sign: 𒊬  SAR — 26 reading values: e₁₄ kiri₆ ma₄ mu₂ mud₆ … and 14 more
urn:nabu:etcsl:0.2.06:31 [sux]
  cubur-e cu [mu2]

Writing systems are a search cut (--script ogam for the Ogham stones, --script grek for Greek-alphabet Gaulish), and so is geography — --within 59.86,17.64,30 walks the runestones raised within 30 km of Uppsala. Every one of these composes with the date window, genre facets, license classes, and lect filters above.

If your field is not here

These walk-throughs sample the disciplines the library serves; they are not the whole of it. The desk index at Research axes covers all 25 — the Indologist, the Koreanist and the Tibetologist among those with no walk-through above — each with its own member shelves, instruments and CLI recipes.