As with any hypothetical design, one is limited to consider if it's making some "unreasonable" assumptions.
First, one may compare the claimed performance to existing technology. Nuclear rockets were abandoned early, but the completed prototypes give some lower bound for what's possible (NERVA, RD-0410).
- Thrust: 2000 kN (NERVA 247 kN, RD-0410 35 kN)
- Isp: 1200s (NERVA 840s, RD-0410 910s)
- Mass: 40 tons (NERVA 18 tons, RD-0410 2 tons).
This isn't wishfully far from demonstrated capability, and importantly, it doesn't assume the increased thrust and exhaust velocity can be achieved on an equal mass budget, instead listing the system as bulkier.
As for "much greater thrust and less radiators (and less dependence on alpha) than a typical nuclear electric rocket", this is conceptually possible, as waste heat can be dumped into the hydrogen instead of requiring additional radiators. The obvious cost of the increased thrust and reduced cooling requirements is wastly inferior exhaust velocity compared to other electrical propulsion types.
The core idea, however, consists of two parts with wildly different technological readiness. Separating reactor from propulsion.
- The reactor, being separate, can use conventional reactor technology, as thus has drastically fewer design constraints. This looks like a massive realism bonus to me.
- This makes an "arc heater" necessary.
2.) Is problematic. It's essentially the "magic device" that enables the whole system, and the author spends a single paragraph describing it!
The electrical energy produced by the latter is then used to power
arc heaters in the exhaust system. This augmentation heating of
the hydrogen raises the effective exhaust velocity well beyond
the 8 km/sec that could be expected from a conventional nuclear thermal engine, to above 12.5 km/sec
No details whatsoever, except that it's somehow better than nuclear thermal engines.
As far as I know, nothing even close to this kind of gigawatt scale electrical heater has been demonstrated for rocket propulsion. Not surprising, since the current available power in space is typically 5 to 6 orders of magnitudes lower.
The available literature about this magic space heater doesn't seem to be large:
Scorpion’s main propulsion system is a version of the Serpent
engine cycle devised by Alan Bond. The cycle uses electrical
augmentation of a nuclear thermodynamic rocket; a basic concept that goes back to at least Goldsmith in 1959 . The third
edition of Sutton  devotes a paragraph to the idea, citing
Goldsmith, and also Berry  and Rester and Rott . Later
editions of Sutton removed this paragraph, presumably to make
room for more detailed discussion of the NERVA development
work that had been undertaken between editions. But the concept was not forgotten and in 1972 Bond outlined a detailed,
elegant and realisable variant on this basic principle 
- M. Goldsmith, “Augmentation of Nuclear Rocket Specific Impulse
through Mechanical-Electric Means,” American Rocket Society Journal,
Vol 29-8, pp 600-601, 1959.
- G. P. Sutton, Rocket Propulsion Elements, 3rd Edition, John Wiley, 1963.
- E. R. Berry, “Effects of Electrical Augmentation of Nuclear Rocket Flight
Performance,” American Rocket Society Journal. Vol 31-1, pp 92-94 1961.
- E.L. Resler and N. Rott, “On Rocket Propulsion with Nuclear Power”,
American Rocket Society Paper 1201-60.
- A. Bond, “A Nuclear Rocket for the Space Tug”, JBIS. Vol 25 pp 625-641,
A paragraph that got removed in later editions, and small stubs at 2-3 pages is not much to go after. Perhaps Bond's outline contains something? Because there's is nothing of substance this paper about how the engine works, only what kind of spaceship you could build around such an engine.
The question is whether one should believe the author.
About the engine's potential applied use? Perhaps, at least he's made some pages about it.
About the engine's design, workings, theoretical foundation, parameters and performance? Not so much, since there's almost nothing said about it.