s-qa
S QA
Establish the deliverable under review from the supplied acceptance and candidate evidence before choosing or loading specialist checks.
Protect the product and improve the next decision. Inspect a frozen candidate without modifying maintained product or maintained tests. Temporary probes and evidence may live outside maintained output for the check. Bind every observation to the exact candidate; a product mutation invalidates affected evidence.
Independent acceptance starts in a fresh context with the accepted contract, candidate identity, and relevant project constraints, without inherited builder conversation or rationale. A review performed in the builder's context remains self-verification.
Select the evidence depth
- Fast: focused feedback near a mutation unit. Check the changed behavior, core journey, applicable acceptance, boundaries, and silent wrong output. Exclude simulations and known long-running checks. Fast QA cannot claim final acceptance.
- Slice: exercise a meaningful vertical slice across its real layers and interfaces. Use rendered interaction or simulation when that is where the failure can exist.
- Final: exercise accumulated acceptance, cross-slice behavior, regressions, integration, and the destination as a whole. This mode owns justified end-to-end simulations and reuses current unaffected evidence.
The project's testing pyramid and defined checks remain authoritative. These modes describe feedback placement and cost, not code-versus-no-code techniques.
Produce actionable evidence
For every applicable acceptance condition, record the action or command, observed result, evidence location, and pass/fail/blocked verdict. Add a verifier-chosen probe against the riskiest plausible false pass.