foundations-distributed-systems
Installation
SKILL.md
Distributed Systems Foundations
When to Apply
Apply when:
- 2+ nodes participate in a write or shared state (replication, consensus).
- A retry, timeout or lost response can duplicate or drop an effect (idempotency, "exactly once").
- Two nodes may both act as primary (leases, fencing).
- A consistency level is being chosen per feature (linearizable / causal / eventual).
- Multi-AZ or multi-region failover and quorum sizing are being decided.
Skip, or route elsewhere, when:
- No coordination, retry/ambiguous-completion, durability or causal-order question exists. A single node or single writer alone is not a skip condition: a committed effect followed by a lost response can still be duplicated by a serial retry.
- The question is throughput or latency under load: use foundations-queueing-theory.
- The question is overload that persists after the trigger, retry storms or load shedding: use qa-resilience.
- The question is availability or SLO budget: use foundations-reliability-theory.
- The question is which verification tool to use (TLA+, model checking, proofs): use foundations-formal-methods.
- "Strong consistency" is a vague preference, not a stated business invariant: challenge it first; the cost is high.