review-gatelisted

# Review Gate Use this after implementation and before merge. ## Division of Labor - **Mechanical checks — owned by automated review (e.g. Alibaba Code Review) and CI**: linting, formatting, conventional style, and straightforward static-analysis findings. When automated review tools like Alibaba Code Review are configured, their output is recorded via `ai-agent-pr-metadata`. - **Semantic review — owned by this gate**: requirements compliance, correctness, regressions, security/authorization, contract/integration risk, and acceptance-test adequacy — everything mechanical tooling can't judge. ## Workflow 1. Confirm linked issue and acceptance criteria. 2. Confirm verification commands and results. 3. Read Alibaba Code Review's findings and the GitHub CI/check results on the PR before starting the independent review — don't re-derive what's already there. If no automated code review tool is configured on the repository, rely on CI check results and proceed directly with independent semantic review. 4. Review spec compliance before code style. 5. Check business logic and edge cases, regressions and compatibility, security and authorization, API/database/queue/integration contracts, and test adequacy including missing acceptance tests (see [canonical reviewer scope checklist](references/reviewer-scope-checklist.md), which already covers backwards-compatibility explicitly). 6. Revisit an Alibaba Code Review or CI finding only when it indicates an unresolved correctness, secur