validate-profile

SUSPICIOUS. The core purpose is coherent for a validation skill, but its actual footprint extends into executing unverifiable local helper scripts and probing arbitrary MCP endpoints defined in local config. That is proportionate to health checking, yet the data flow trust boundary is weak: endpoint domains are unconstrained, and the local scripts could handle credentials internally in ways not visible here. No confirmed malware, credential theft directive, or malicious installer is present, but the network and helper-script trust model makes this a medium-risk skill.