Teoria

The core loop is a mapping from declared risk to reproducible evidence.

Finding=Oracle(Probe(Scenario(Invariant(Risk)))) Finding = Oracle(Probe(Scenario(Invariant(Risk))))

The useful quality signal is not the volume of generated scenarios. It is the ratio of verified findings to total explored risk surface.

Q=FvRe+C Q = \frac{F_v}{R_e + C}

Where F_v is verified findings, R_e is explored risk area, and C is execution cost.

Optimize for evidence

Better QA runs reduce ambiguity. They do not merely produce more prose.

Practical implications

  • A high-severity risk with no invariant is underspecified.
  • A scenario with no oracle is exploration, not verification.
  • A finding with no replay is a lead, not a release-gate blocker.