generated-test-self-reviewlisted

# Generated Test Self-Review > **Standalone adaptation.** This is a self-contained version of the self-review lint that ships > with the `automate` skill of [QAIA](https://github.com/QAIA-Project/QAIA) (MIT). The canonical > version, the before/after fixture that demonstrates it, and the five judge runs that produced > classes D5-D9 live in that repository. QAIA is pre-alpha and says so. ## When to use this **Before a generated spec file is written to disk**, not after. Every defect below costs one line to fix during generation; the same defect caught later costs a review cycle, and caught by nobody costs the credibility of every green run in the suite. This is a **proofread inside generation**, not a score and not a gate. ## The two families **D1-D4 are shapes**: an assertion that cannot fail whatever the application does. A static reader catches them. **D5-D9 are invisible to that reading.** Every one is a *real* assertion, on real state, that any static tool accepts. They were found by five independent judges reading five generated suites against their specifications — **none of the five passed the quality gate**. --- ## D1 — Tautological or reflexive comparisons `expect(true).toBe(true)`, `expect(1).toBe(1)`, `expect(x).toBe(x)`. A constant asserted against itself. No application state is involved. ## D2 — Contentless `expect()` calls `expect(true).toBeTruthy()`, `expect("ok").toBeTruthy()` — an argument that is a hardcoded literal rather than something read