cloud-design-patternslisted
Install: claude install-skill sandbaseai/workbuddy-skill
# Cloud Design Patterns
## Purpose and boundary
Use patterns as named hypotheses about a recurring problem, not as proof that a design is correct. A pattern can add latency, cost, state, operational burden, or new failure modes. Keep analysis and plans read-only until the owner authorizes implementation; this Skill does not provision infrastructure, change routing, migrate data, or alter access policy.
## Start with constraints
Capture the workload, critical user journeys, dependencies, data flows, trust boundaries, traffic shape, latency, availability, recovery, compliance, data residency, cost, team ownership, and provider limits. Define measurable fitness functions such as p99 latency, error rate, queue age, recovery time, data loss, unit cost, or operator effort.
Ask before choosing a pattern:
- What failure or change is the pattern meant to contain?
- What is the authoritative state and who owns it?
- Is the operation synchronous, asynchronous, replayable, idempotent, or compensatable?
- What happens under timeout, duplication, reordering, overload, stale data, partition, and partial deployment?
- What new credentials, network paths, storage, logs, queues, retries, or on-call work does it create?
- How will the decision be tested, rolled back, retired, and reviewed?
## Pattern selection workflow
1. **Frame the problem.** Describe the user or system outcome and the current failure mode; do not start with a favorite pattern.
2. **Map the boundary.** Identify caller