| name | e2e-verification |
| description | Activate this skill when verifying changes, running E2E tests, or checking DoD requirements before PR creation. |
E2E Verification & Definition of Done
Before any feature, bugfix, or issue can be considered "code complete", the agent must ensure that all integration and verification gates pass.
1. Automated E2E Testing
Run the local E2E Playwright test suite against a real Liferay DXP container:
bash scripts/run-e2e-ldm.sh -v -k --ci
Before declaring any feature or bugfix 'code complete', you MUST execute run_command to invoke the command above, and END your turn. You are FORBIDDEN from declaring the task finished or pushing PRs until you receive the output confirming the E2E verification succeeded.
2. Post-Completion "Definition of Done"
- Test Protocol: Provide 3-5 manual/automated steps to verify in a live Liferay instance.
- Redundancy Scan: After a feature is complete, scan for any newly introduced duplicate code.
- Strategic Deployment Control (No Automatic Deploy): Do not run or suggest
deploy tasks as part of a general "build" command.
- Dependency Awareness: Before deployment, list the required order of execution (e.g., 1. OAuth2 CX, 2. Batch CX for Objects, 3. Frontend Custom Element).
- Manual Trigger: Always end a feature cycle by asking: "The code is ready and tested. Would you like me to provide the specific build/deploy commands for this extension now?"
Last Updated: 2026-07-19 | Last Reviewed: 2026-07-19