Work out mutation score instantly with clear inputs, formula shown and shareable results.
Mutation testing injects small behavioural changes and checks whether the suite notices. The score is killed mutants over viable ones, excluding equivalent mutants that cannot change observable behaviour. The gap against line coverage is the interesting number: a large gap means tests execute the code without asserting on it, which is the failure mode line coverage cannot see.
Mutation score
viable = generated - equivalent; score = killed / viable; gap = line coverage - mutation score
The suite must run once per mutant. Practical tools mitigate this with coverage-guided selection so only tests touching the mutated line are run.
Above 70 percent on core business logic is strong. Applying it to the whole codebase is usually too slow, so scope it to the modules where correctness matters most.