Test strategy

Decide which risks the suite must catch, which kind of test catches each one, and what judges a result correct. A strategy is a table of risks, not a pyramid quota or a coverage percentage.

Process

1. Inventory. Read the test config, runners, scripts, fixtures, helpers, CI workflows and any testing guide. Record what runs on every change, what runs on a schedule and what only runs by hand. A suite nobody runs is not coverage.
2. List the risks. For the change or the product area: wrong numbers, wrong permissions, invalid input reaching storage, broken flows, contract drift between services, slow or chatty screens, inaccessible controls, visual regressions, unsafe rendering, model output regressions, convention drift. Name the observable symptom of each.
3. Assign a test type and layer per risk. Route each to the skill that owns it: