Convert validated locators into typed elements, primitive action methods,
and high-confidence composite actions. This sprint creates the bridge
between locator intelligence and safe page-object / test-skeleton generation.
Validated locators can be typed into meaningful UI intents
Primitive methods can be generated deterministically
Composite workflows can be synthesized only when confidence is high
Bounded generation is safer than full test guessing
Core output chain
locator-map.json
element-type-map.json
action-method-candidates.json
composite-action-candidates.json
What gets generated
Primitive actions
fillUsername(value)
fillPassword(value)
submitLogin()
clickPublish()
Composite actions
loginAs(username, password)
Only when intent evidence is strong
Only when method confidence clears threshold
Why this matters
Most automation systems either stop at locator discovery or jump too quickly
into unsafe full-test generation. This sprint takes the middle path:
turn validated system understanding into reusable methods without pretending
the system knows more than it actually does.
✔ less boilerplate page-object code
✔ stronger consistency in generated methods
✔ confidence-gated composition
⚠ uncertain flows remain reviewable instead of over-generated
Who this is for
Teams exploring controlled AI in QA without surrendering determinism
Teams that already have validated locator artifacts
Teams that want a safer bridge into page-object and test generation