Agent Skills: System Architecture

>-

UncategorizedID: jagreehal/jagreehal-claude-skills/system-architecture

Install this agent skill to your local

pnpm dlx add-skill https://github.com/jagreehal/jagreehal-claude-skills/tree/HEAD/skills/system-architecture

Skill Files

Browse the full folder contents for system-architecture.

Download Skill

Loading file tree…

skills/system-architecture/SKILL.md

Skill Metadata

Name
system-architecture
Description
>-

System Architecture

Critical rules

  • No universals — only trade-offs in context. Prefer the simplest design that meets current needs and enables growth.
  • Always present 2–3 valid approaches; analyze gains and losses explicitly.
  • Record expensive-to-reverse decisions as ADRs; never delete historical ADRs.
  • Earn complexity (microservices, sharding) when boundaries and scale demand it — not because a famous company did.
  • Before choosing patterns or writing ADRs, read references/patterns.md and references/adr-template.md.

Workflow

  1. Understand context: business needs, team skills, time/budget constraints.
  2. Identify 2–3 approaches; fill a trade-off table (see references/patterns.md).
  3. Check what is hard to change later and what matches team capabilities.
  4. Decide and document with the ADR template in references/adr-template.md.
  5. Define boundaries/contracts so teams can move independently.
  6. Revisit when team size, scale, or bounded contexts materially change.

Resources

Validation

  • [ ] Context understood before recommending
  • [ ] Multiple approaches identified with explicit trade-offs
  • [ ] Long-term change cost considered
  • [ ] ADR written for significant decisions
  • [ ] Matches team capabilities; simplest workable solution

Constraints

  • Do not design for nonexistent problems or write ADRs for trivial reversible choices.
  • Related: documentation-standards, design-principles, api-design, resilience, data-visualization.