Phase 5 — Quality & Planning
Test Strategy
The test plan that makes AI-generated code trustworthy
2 min read · Document 20 of 30
What is this document?
The Test Strategy defines what behaviour matters and how it is verified at each level — unit, integration and end-to-end — with the tools, environments and test data rules.
It is written before tests exist, so tests verify behaviour instead of mirroring the implementation.
Why does it matter?
- AI tools generate tests that test implementation, not behaviour.
- The pyramid keeps the suite fast and reliable.
- Test data rules keep production safe from test runs.
What's inside?
- Test levels — What each level covers and how much.
- Environments and data — Where tests run and how test data is isolated.
- Critical paths — Journeys that must have end-to-end coverage.
UnitIntegrationEnd-to-end
How CREAH generates it
CREAH builds this document from your Stage 0 answers and interview. Questions that shape it include:
- Which journeys would hurt most if they broke?
- Is there a staging environment?
- Can tests use real payment sandboxes?
Example excerpt
## Level 3 — End-to-end
Scope: the 5 critical journeys only (signup, pay, interview, generate, download)
Environment: localhost or staging — refuses to run against production
Data: every test row carries is_test = TRUE and is removed after the run