It's difficult to say much about Khuzdul verbs with any certainty.  The only examples we have are the stems felek "to hew", gunud "to tunnel, excavate", and salôn / sulûn "to fall, decend swiftly".  We are also told that felak "hewer" can also be used as a verb meaning "to use a hewer".  Because Tolkien made mention of Khuzdul's Semitic structure on several occasions, we can maybe infer that the verbal system resembles that of Hebrew or Arabic.  It would certainly make it much different than the other languages in Middle-earth.   If so, we can view several other roots as being verbal in nature: '-G-L "to speak", G-B-L "to be great, mighty", G-M-L "to be/grow old, to age", Z-R-B "to record", Sh-R-B "to be/become bald", S-G-N "to be long", L-Ch-R "to smith", Z-B-D "to rule", and Z-H-R "to be hollow".  Note that a number of these are very hypothetical, so refer to the pages of the various words that contain them to see why I assign these meanings.

When viewed in total, these roots can be analyzed as being very similar to Arabic's verbal system.  Doing so actually makes them seem like a coherent whole.  There are several features that lead me to this conclusion.
  • The roots G-B-L "to be great, mighty", G-M-L "to be/grow old, to age", Sh-R-B "to be/become bald", S-G-N "to be long", and Z-H-R "to be hollow" can all be viewed as "stative verbs", which indicate a state the subject is in as opposed to an action being performed.
  • The words gabil "great" and gamil "old" are especially similar to the forms of verbal adjectives of stative verbs in Arabic.  The template for these words in Arabic is CaCîC, which we can easily view as being reduced to a CaCiC composition form in Khuzdul.
  • The words felak "hewer" and uzbad "ruler" can be viewed as active participles, even though they differ in vocalization from the standard template for active participles in Arabic, which is CâCiC.  Arabic participles often become lexicalized, which would be the case here.  We could view these words as the "hewing" and "ruling".  Felak especially fits the view of an active participle since it is a tool, but doesn't have a template used for tools in Arabic: miCCaC.  This too fits Arabic, as the use of the miCCaC pattern isn't a universal for implements.  Sometimes the active participle is employed.  This is an example of where the active participle becomes lexicalized.
  • Mazarb can easily be seen as a passive participle.  The passive participle in Arabic typically takes the template maCCûC, which is a bit different than mazarb.  Still, we see the ma- prefix here, which is seen in other Semitic languages, so mazarb fits well as a passive participle.
  • The only way I was able to interpret iglishmêk was to view 'igl as a gerund (or "verbal noun").   The pattern CiCC is used in Arabic for the gerund (called a "masdar") of a verb.  Hebrew and Yiddish use infinitives and participles, and those structures just didn't fit the evidence in Khuzdul quite as well as the Arabic participles and gerund.  A gerund similar to Arabic also provides an explanation for why we see the pattern CaCC in Sharbhund as opposed to CaCaC or CaCiC.

If we accept that many Khuzdul words can be viewed as fitting a structure similar to Arabic and perhaps incorporating elements of other related languages (Yiddish, Hebrew, and perhaps other Semitic languages), then there are a couple of major features that we can probably expect to see.
  • Both Arabic and Hebrew verbs, especially in ancient times, have two prime conjugation forms: the perfect and the imperfect.  The perfect is primarily formed with suffixes, while the imperfect is formed with prefixes.  As the base conjugations, they also make these languages primarily aspect driven rather than tense driven.  This is a very different paradigm than Indo-European languages.  (It should be noted that the claim of being primarily aspect-based rather than tense is debated by scholars.)  If Khuzdul follows this structure, then it would also be much different from the other languages found in Middle-earth, and would add to the view by non-native speakers of it being a complex language.
  • Arabic verbs have ten to fifteen "forms" or "structures" which modify the meaning of the verb in some way.  Form I is generally the most basic form and meaning of a root, and other forms will take on various interpretations, such as the causative or reflexive.  Hebrew has a similar concept, which is the seven "binyanim", or again "structures".  In Hebrew, the meanings of each binyam seem to be more standardized than the Arabic forms.  On the other hand, each Arabic form is more standardized in its template or vocalization than in Hebrew.  Still, the concepts are somewhat similar and both provide the language with a way to derive various verbal meanings from the same root.

There are other features of Hebrew and Arabic that are similar and could very possibly show up in Khuzdul, but the perfect/imperfect aspects and the forms/binyanim are so central to Semitic languages that it's difficult to see Khuzdul without them and still be called "Semitic" in nature.

There is one manner in which Khuzdul verbs are apparently distinct from Hebrew and Arabic.  The vowel patterns seen in Khuzdul verbs seem to vary quite a bit.  As above, we have the examples felek, gunud, salôn/sulûn, and felak.  Each of these would probably be, or be similar to, the most basic forms of the verbs.  By contrast, the basic forms in Hebrew and Arabic are relatively set into a single form by the "Pa'al" and Form I conjugations respectively.  If Khuzdul verbs take on many different base forms and still have a binyanim/form architecture on top of those forms, it could make Khuzdul incredibly complex.