Liv(y) in LSJ

All of the citations in LSJ of Greek words from Livy can be automatically collated (complete in theory).  

Liv.Ann.3 is vol. 3 of Liverpool Annals of Archaeology and Anthropology, 1908-. Its inscription for διαυλόδρομος has a PHI text, as does its citation for Περφερέτας (prob.)

No real surprises: military technology, festivals, flora, etc. There is a kind of coin (mentioned also in Cicero's letters). τυραννίς II 3 is noteworthy.

LSJ and Catullus and Martial

Catull(us) is cited four times in LSJ: once as one of the earliest attestions of παλίμψεστος (with Cicero), once for a new meaning of (!) φάσηλος (with Horace), once for (etymologically correct) aspiration of Ionic ἀπηλιώτης (cf. Helios, sol), and once for explanatory purposes s.v. σκορπίος.

LSJ's 'Authors and Works' does not include 'Catull.' (p. xxi), but the Supplement (p. vii) adds 'Cat.', which remains uncorrected in the Revised Supplement (p. xii).

Mart(ial) seems to be cited more frequently (but the following list has not been reliably created): s.vv. ἀνακλινοπάλη ('Mart.14.201 (f.l. for ἐπι- )'), δρῶπαξ, ἐπιδειπνίς, ἐπικλινοπάλη, κορδύλη,  λῆμμα, λύγδος, παλίνουρος ('with play on pr. n. Palinurus [the individual in books 3, 5, and 6 of the Aeneid], Mart.3.78.2'), ῥῠτός (actually, for ῥύτιον),  σίγμα (cf. s.v. Σ), σοφός, σύνθεσις, and ψυχή.

'Mart.Cap.' and 'h.Mart.' account for most other instances of 'Mart.'.

A Greek word known from the Greek of a Latin speaker

Marcus Aurelius has κοινονοημοσύνη (1.16[.2]), which is a hapax in the TLG. Several Latin authors use ἀκοινονόητος, which is not found in any Greek text in the TLG. LSJ translates aptly through the medium of French lacking in 'savoir-faire' (Lat. sensus communis).

Latin nouns in -a-ster

Augustine, City of God II.27 begins:
vir grauis et philosophaster Tullius aedilis futurus clamat in auribus ciuitatis...

-aster has been added to the stem philosoph-a-- (cf. philosophari, etc.).

Herein is Augustine's verdict on Marcus Tullius Cicero. He has in mind In Verrem 2.5[14.]36 and soon afterwards he refers to In Catilinam III [8.]20.

oleaster tri 'the wild olive' can be cited from Cicero, Sallust, Virgil, and Ovid. Whence oleastellus 'A cultivated variety of olive resembling the wild sort in its fruit' Col(umela) 12.51.3.

OLD's entry for this suffix (p. 191) also cites parasitaster (Terence, Adelpi 779). Cf. parasitari

Also of note is the name of the Pauline commentator Ambrosiaster. Apparently, this name 'Would<->be-Ambrose (perh. 'Ambrose-Wannabe') is not antique, but early modern.

adiutare in a fragment of Cicero

In 'Petronius and the New Non-Literary Latin', J.N. Adams refers (p. 12, n. 7) to a single occurrence of the frequentative adiutare (contrast adiuuare) 'in a letter'. He does not give a precise citation, but refers to p. 80 of his 1977 work on Claudius Terentianus.

The instance in Cicero is ep.fr.11(10)3, according to OLD, which cites P.Mich.471.28 (Claudius Terentianus: images), Lucr.1.812 and Sallust's epistle de re publica 2.12.4, as well as Plautus and Terence, as mentioned by Adams.

This fragment can be found in the OCT M. Tulli Ciceronis: Epistulae, III (1902 ed. L.C. Purser [not paginated] and 1958 ed. W.S. Watt, p. 171). Arusianius Messius vii, p. 453.23 noted (on adiuta hoc illos) Cicero Ad Axium, ex libro I: si tu aliquid nos adiutare potes.

Greek words known exclusively or primarily from the Scriptores Historiae Augustae

LSJ has the sequence 'Hist.Aug. ...' -

As the only citation in the following entries (all in Latin script; variants and mss. spellings, etc. noted):

ὀξυπαιδερώτινος Hist.Aug.Aurelian.46<.4> [acc.pl. in the context: oxypederotinas with reflecting the monophongisation of the diphthong formerly written ]. 

πανδουρίζω, play the πανδοῦρα, Hist.Aug.Elag.32<.8>:  
actually pandurizavit (but, cf. πανδουριστής, οῦ, ὁ, one who plays it, Euph. Fr.Hist. 8, JRS 18.177 (Jerash)')

πενταφάρμακος Hist.Aug.Hel.5 [= Spart. Ael. Ver. 5<.4 and 5> Lewis and Short, s.v.: pentefarmacum]. tetrapharmacum also appears 2x in the context [terrafarmacu P^1 pentefarmacum P pentafarmacum Pet post pentefarmacum add. Σ ut alii ferunt & pentafarmacu P terrafarmacu P^1]

χρηστολόγος giving fair words, speaking plausibly, Hist.Aug.Pertinax 13.5 [christologum P (chestologon Epi. de Caes. 18, 4)].

As one of two citations for an entry: αὐθέψης Cic.Rosc.Amer.46.133, Hist.Aug.Elag.19<.3> [autepsas P mensas et capsas superscr. a. m.].
Russian samovar mirrors the structure of the word (illustrations), but, presumably, is not a calque.

Dyck comments on the Cicero passage, as follows: 
"An authepsas was a cooker containing its own heating element, the Romans, as often, having borrowed the name along with the object from the Greeks; our passage is the word's first and only attestation [we could add 'in either language']; cf. OLD [which does not extend its coverage beyond Apuleius] and TLL s.v.". He goes on to quote Seneca, Ep.78.23, where this kind of device is in view, but this noun is not used.

Neither the OCT nor the Teubner report any variation in the spelling of this word.

LSJ does not have Ros.Amer. in its Abbreviations list under Cic. Only works of rhetorical theory and philosophy are mentioned there, as well as the collections of letters (unsurprisingly from the point of view of code-switching and lexical borrowing). A full list of instances of 'Cic.' in LSJ is easily generated.

As a (unique) citation for a subsection: ἀγαθοδαίμων II Hist.Aug.Elag.28<.3>.