Case category · R&D & quality
5 cases Category 1 of 20
This band covers the everyday engineering loop before and after merge: human- or agent-assisted review checklists, test and coverage backfill, third-party dependency and license risk, keeping API contracts and docs aligned, and phased stack migrations with sign-off. When you orchestrate skills from the skills library—code review, testing, compliance, API—use these cases as trigger scenarios and delivery templates.
In the case hub, this is the R&D & quality band in the delivery → data → security → growth → general layout (anchor #cat-dev). Each case page states when to trigger, what to deliver, and what “done” means; entries are examples you can merge or split for your team.
Review checklists, risk hotspots, merge gates, and CI alignment.
Backfill tests, edge cases, coverage targets, and gap lists.
Dependency tree, license conflicts, compliance reports, upgrade guidance.
OpenAPI/contract changes, docs vs implementation consistency.
Migration paths, compat layers, rollback points, phased sign-off.
Structure review focus as a checkable list: security-sensitive changes, concurrency and transactions, error handling, observability hooks, feature flags when needed; align recommendations with CI gates and branching policy, and flag items that always need a human.
Locate gap files and branches from existing coverage reports; add edge and failure-path tests; state team targets (line/branch) and explicit exceptions, producing a prioritized backfill list you can review in retros.
Resolve the dependency tree and transitive packages; flag licenses that conflict with commercial policy (e.g. GPL/AGPL); summarize upgrade paths, alternate packages, and lawyer-ready conclusions for release or procurement.
Diff implementations against OpenAPI/Protobuf contracts and doc sites: deprecation, example requests, error catalogs; after changes, ship a sync checklist and impact notes for web and mobile consumers.
Phase the migration (dual-write, read cutover, retire old stack), design compat shims and feature flags; define sign-off per phase, rollback criteria, and monitoring so you avoid a single big-bang cutover.