| 1 | Requirements: FRs, NFRs, constraints | Define required behavior, quality targets, constraints, assumptions, and proof. |
| 2 | Boundaries, state & data | Know ownership, source of truth, dependencies, and failure boundaries. |
| 2A | Networking & communication | Choose protocols, connection style, geography, and network failure behavior intentionally. |
| 2B | Data modeling, indexing & partitioning | Model from access patterns; optimize before sharding; choose partition keys carefully. |
| 2C | Time, clocks & ordering | Make time-zone, expiry, and ordering semantics explicit. |
| 3 | Concurrency | Protect shared state and reason about simultaneous work. |
| 4 | Transactions & consistency | Choose transaction boundaries and the consistency model intentionally. |
| 5 | APIs & idempotency | Treat contracts, retries, versioning, rate limits, and duplicate requests as design concerns. |
| 6 | Messaging & asynchronous work | Design for duplicate delivery, ordering, backlog, and poison messages. |
| 6A | Real-time & long-running work | Choose polling, SSE/WebSockets, or background jobs based on actual interaction needs. |
| 7 | Failures & resilience | Plan timeouts, retries, circuit breaking, degradation, rate limiting, and recovery. |
| 8 | Scale, performance & caching | Estimate load, find bottlenecks, bound resources, and control stale data. |
| 9 | Security | Define identity, authorization, trust boundaries, secrets, and data protection. |
| 10 | Observability & reliability | Make behavior measurable with telemetry, SLOs, health, backup, and recovery. |
| 11 | Deployment & evolution | Change the system without breaking clients, data, or production. |
| 12 | Cost, simplicity & operability | Avoid architecture that is more expensive or complex than the problem. |
| 13 | Master checklist | A reusable end-to-end design review. |
| 14 | Requirements-to-delivery lifecycle | Carry FRs, NFRs, constraints, ADRs, TIP, implementation, verification, and production feedback. |
| 15 | LLM and agentic systems | Design model, context, retrieval, tools, memory, orchestration, safety, evaluation, and operations. |
| 16 | Spec-driven development for agents | Turn intent into versioned behavior, tool, context, evaluation, implementation, and release contracts. |
| 17 | Agent-system review checklist | Review autonomy, protocols, security, reliability, observability, cost, and human control end to end. |