Comprehensive reference for system design across UI design systems, frontend, mobile, backend HLD, low-level design, SRE, SaaS, and migration. Use this skill for architecture decisions, scaling, database selection, RAG, agents, observability, and cost optimization. Trigger on: "how should I design X", "which is better X or Y", "best practice for Z", "edge case in X", "how to handle X at scale", "is X deprecated". Always follow the user's preferred language for communication while referencing English knowledge.

System Design Overview — Reference Skill

Full Landscape

┌──────────────────────────────────────────────────────────┐
│                      Client Layer                         │
│  ┌─────────────┐  ┌─────────────┐  ┌──────────────────┐  │
│  │ UI Design   │  │ Frontend    │  │ Mobile App       │  │
│  │ System      │  │ Web         │  │ iOS / Android    │  │
│  └─────────────┘  └─────────────┘  └──────────────────┘  │
└──────────────────────────────────────────────────────────┘
┌──────────────────────────────────────────────────────────┐
│                      Server Layer                         │
│  ┌───────────────────────────────────────────────────┐   │
│  │        High-Level Design (HLD) — Infrastructure   │   │
│  └───────────────────────────────────────────────────┘   │
│  ┌───────────────────────────────────────────────────┐   │
│  │        Low-Level Design (LLD) — Implementation    │   │
│  └───────────────────────────────────────────────────┘   │
└──────────────────────────────────────────────────────────┘
┌──────────────────────────────────────────────────────────┐
│     Cross-Cutting: Security · Observability · CI/CD       │
└──────────────────────────────────────────────────────────┘

Routing — Which file to read?

Domain	File	Trigger
Deployment & Release	`references/deployment-release.md`	"safe deploy", "CI/CD", "canary release", "blue green", "rollback"
SaaS & Multi-tenancy	`references/auth-multi-tenancy.md`	"SaaS design", "authz", "auth", "tenant", "multi-tenant"
SRE & Incident	`references/sre-incident-response.md`	"system down", "observability", "alert", "post-mortem", "SLO/SLA"
Edge Cases	`references/edge-case-analysis.md`	"edge case", "system failure", "race condition", "failure mode"
Diagrams	`references/documentation-diagrams.md`	"draw X diagram", "Mermaid", "C4 model", "Sequence diagram"
Migration	`references/migration-strategy.md`	"modernize system", "switch to microservices", "migrate database"
ADR (Records)	`references/adr-guide.md`	"record decision", "ADR", "why choose X?", "architecture logging"
Tech Selection	`references/tech-selection-strategy.md`	"which tool?", "compare X and Y", "modern stack for Z"
Quick decisions	`references/decision-trees.md`	"use X or Y?", "what to choose?"
Anti-patterns	`references/anti-patterns.md`	"should I use X?", "why is X failing?"
Sizing & numbers	`references/sizing-guide.md`	latency, throughput, thresholds, cost
AI Engineering	`references/ai-engineering.md`	RAG, LLM, agents, MCP, LLMOps, vector DB
Backend / HLD	`references/backend-hld.md`	API, DB, cache, queue, scale
Low-level Design	`references/lld.md`	patterns, CQRS, data model, algorithms
Testing	`references/testing-fundamentals.md`	Unit, integration, E2E, contract, performance
Data Pipelines	`references/data-pipelines.md`	ETL/ELT, CDC, Spark, dbt, Iceberg
Compliance	`references/compliance.md`	GDPR, HIPAA, PCI-DSS, PII
Edge & WASM	`references/edge-wasm.md`	Cloudflare Workers, WebAssembly, WASI
FinOps	`references/finops.md`	Cloud cost, unit economics, reserved instances
Cross-cutting	`references/cross-cutting.md`	Security, observability, CI/CD, SRE

Read decision-trees.md FIRST if the user is choosing between options. Read anti-patterns.md if the user asks about a specific approach. Read multiple files if the question spans multiple domains.

3 First Questions (Every System)

Scale — DAU? QPS? Read-heavy or write-heavy? Real-time or batch?
Team — Size? Monolith or microservices? Web-only or mobile as well?
Constraints — Deadline, budget, existing stack, compliance (GDPR, PCI)?

Latency benchmarks (Memorize these)

L1 cache:        ~1 ns
RAM:             ~100 ns
SSD read:        ~100 µs    (1000× RAM)
Network same DC: ~500 µs
HDD seek:        ~10 ms     (100× SSD)
Network cross DC:~100 ms

Cache hit vs DB query: ~1000× | RAM vs disk: ~100×

Checklist severity convention (Used throughout the skill)

🔴 MUST     — Block ship if missing, security risk or data loss
🟠 SHOULD   — Fix before production, acceptable with documented exception
🟡 NICE     — Tech debt, minor DX/performance impact, fix over time