Checklists

Objective, measurable, automation-friendly review gates. Have an agent create a todo per item and self-review before declaring work done. Each maps to a standard and/or an agent.

• architecture_review.md — layering, DI, modularity
• code_review.md — correctness, security, tests, standards (baseline gate)
• uikit_review.md — programmatic views, MVP contract, presenter rules, lifecycle, accessibility
• security_review.md — OWASP MASVS storage/transport/auth/crypto
• api_review.md — REST/GraphQL contracts, errors, pagination
• websocket_review.md — realtime lifecycle, concurrency, backpressure
• performance_review.md — main thread, memory, launch, battery
• accessibility_review.md — VoiceOver, Dynamic Type, contrast
• release_review.md — versioning, signing, store readiness, rollout

Severity convention

Tag findings Critical / High / Medium / Low / Nit. Critical and High block merge/release; Medium/Low may be deferred with an explicit note; Nits are optional.