Lean one-liner definitions ('SoC', 'SSOT', 'Fail Fast') with no padding and no explanation of concepts Claude already knows — every token earns its place.

For a standards skill the guidance is concrete (dependency direction, pattern lists, CAP trade-offs), but it is largely descriptive/definitional rather than executable instructions — matches the 'some concrete guidance but incomplete' anchor.

The 5-step 'Evaluate Architecture' workflow is sequenced, but lacks explicit validation checkpoints or feedback loops; the ADR step is a documentation step rather than a validation gate — matches the 'steps listed but validation gaps' anchor.

SKILL.md is a concise overview with well-signaled one-level-deep links to three real bundle files (distributed-systems.md, resilience-patterns.md, implementation.md), all of which exist and are referenced contextually — matches the 'clear overview with well-signaled one-level-deep references' anchor.

Lists multiple concrete actions ('Enforce separation of concerns, dependency inversion, and resilience patterns') plus design activities (designing features, evaluating module boundaries, selecting patterns, resolving bottlenecks) — closest to the 'multiple specific concrete actions' anchor.

Explicitly states what it does ('Enforce... resilience patterns across layered and distributed architectures') and provides an explicit 'Use when...' clause enumerating triggers, clearly answering both what and when.

Triggers like 'designing new features' and 'evaluating module boundaries' are relevant but lean technical/architectural jargon ('separation of concerns', 'dependency inversion') rather than terms users naturally say, missing common variations — matches the 'some relevant keywords but missing common variations' anchor.

Occupies a clear system-design/architecture niche with distinct triggers unlikely to overlap with unrelated skills — matches the 'clear niche with distinct triggers' anchor.