plant-defects

Deliberately break code to prove a test can see it: plant the defect an instrument claims to catch, confirm it goes red, and record it in a table that regresses. Use whenever the question is whether a check can fail, not only when invoked explicitly: (1) reviewing an assertion or CI gate whose claim is an absence (no leak, no race, no warning, nothing left behind), (2) relying on an external instrument such as a sanitizer, linter, or coverage tool, (3) judging an acceptance criterion that names an instrument it expects to fail, (4) asking "would I know if this broke", or explaining what a green suite establishes, (5) suspecting a check runs on nothing, or that code is unreachable from every test build, (6) choosing between instruments, or deciding not to build one. Covers the plant table, ordered assertions, positive controls, vacuous passes, structural uncoverability, instrument-blindness matrices, and source canaries.