performance

Use when an approved system design exists and the team needs performance architecture before implementation or scale events. Produces explicit latency, throughput, concurrency, and cost budgets per user-visible path, a capacity and headroom model, scaling and backpressure posture, hot-path and critical-path analysis, caching and precomputation strategy, performance test and regression-gating policy, and implementation handoff notes. Do not use for availability and disaster-recovery design (use reliability), telemetry instrumentation (use operations), ad-hoc profiling, single-query tuning, or vendor-specific runtime tuning.