MCAS descriptive wording may not be written in the clearest manner. That aside, I know the design intent and implementation with enough detail to confidently say that MCAS authority is limited to 2.5 degrees of stabilizer (less at higher Mach numbers) and will not add more unless the function is reset. Pilot activation is manual electric switch commanded stabilizer motion is one way to reset MCAS. This appears from the data shown here in PPRUNE to be what was repeatedly occurring during the Lion Air event.
One question that was posed a few days ago is what would have been the MCAS response if the crew were to use manual mechanical nose up stabilizer trim via the cockpit trim wheel following an errant MCAS nose down stabilizer increment. In that case MCAS would not have followed with more stabilizer motion as it’s authority limit would already have been met. I am not suggesting that this would be proper pilot approach to ending the MCAS down, pilot up cycling seen with an erroneously high AOA indication, but mention this scenario by way of providing better system understanding.
On the the question of stall prevention there are a couple of things that clearly reveal that this is not the objective of MCAS. First is it’s limited authority. Clearly generous aft column without pilot pitch trim input would provide much more nose up command that the limited increment of nose down provided by MCAS. The crew can definitely command the airplane into stall in the presence of MCAS. Second is the slow rate of MCAS stabilizer motion. Even running the stabilizer at the relatively quick 0.27 degrees per second it takes almost 10 seconds to insert MCAS’s full increment of stabilizer motion. A sharp column pull would be able to get to a stall before MCAS has a chance to complete its stabilizer motion.