claude-code-architecture-patterns

SUSPICIOUS. The skill is mostly a documentation/tutorial resource, and its examples are proportionate to that purpose, but the install path is inconsistent with the stated role: it asks users to execute code from a personal GitHub repo unrelated to Anthropic and not clearly the same publisher as the skill branding. I do not see credential theft, exfiltration, or malware behavior in the skill content itself; the main risk is third-party supply-chain trust and source/publisher mismatch.