licanfu/multica-best-practices/multica-test-t3-ui-automation

UI Automation (T3 · UI)

T3 only after G2.5 PASS; an AC can only be verified by UI → run this skill alone; if it also has an API channel → run both, orchestration merges one report.

Integrated from ui-automation-squad. Material assembly + BLOCKED see `multica-test-orchestration`.

Required Items (stop + interruption list if missing; summary from comments)

Never use xmind-ui-automation / playwright.git / req-figma-ui-automation as prerequisites.

Flow

pytest <MULTICA.md §2 e2e path> --base-url <page_url> --alluredir ./reports/ui

Cases source: T1 Confluence t1-cases + T2 supplement. Missing script path → interruption list.

Output and merge

# T3 UI Automation — {JIRA_KEY}
-G2.5: PASS @ {sha}  deploy: {url}  page: {page_url}
-e2e: {MULTICA.md §2 path}
-total / pass / fail / error
-per case: id / title / status / screenshot / log path
-conclusion: PASS | FAIL | BLOCKED

multica-test-orchestration merges with API side: one side only verifies a given AC, missing that side → whole ticket cannot be PASS; errors → triage (references/t3-failure-triage.md).

Why it works

UI verification bound to the actual deployed page; 1:1 with cases avoids "runs but doesn't cover"; running after G2.5 avoids a green build masking a broken environment.