There was I thinking that "Project Orion" (ie nuclear pulse jet) was dangerous.
The big reason other than the CTBT why this was stopped was the risk of the small nuclear pulse units (basically itty bitty nuclear warheads) getting into the wrong hands.
Making them on board using berkelium to lower the yield and also make them essentially fail-safe because the half life is so short is doable, in fact I sent some notes about it to NASA and other agencies along with the schematics and calculations.
Essentially my modified design plates about a 250um thick layer of BkFx onto the outside of the normally sub-critical plutonium spheroid, and uses a modification I invented to overcome the need for a bulky and dangerous Po+Be initiator which is the subject of a patent search.
Also because the lasers are what ignites the explosive lenses if something goes wrong nothing happens instead of a fizzle or at worst a <10t yield which is well within the capabilities of a properly designed refractory ejection tube.
The pulse unit I designed is about the size of a coffee cup and despite this has a 220t maximum yield (ie enough for about an Isp of ten times a Saturn V)
Should be enough to get to Mars in about 8 months or less, with the engines running at 25% of maximum power detonating about one pulse unit every 5 minutes with a total fuel usage about equal to 500 kilos of Pu.