If someone can check my math, this seems like a perfect opportunity to do a statistical test. Sample sizes are too small to say that there is a difference between previous infection and no previous infection.

Given the contingency table:

      No prior inf.   Prior inf.  
  Infected   3             0

  Uninfect   4             6

A Fisher Exact Test yields a p value of 0.1923. So there is not evidence that previously infected people are safe. Similarly, if we compare the attack rate in previous infections vs Pfizer, the p value is about 0.15, so there’s not evidence one is better than the other.