tech-decision

SUSPICIOUS. The stated purpose is coherent, but the skill expands its footprint by directing the agent to invoke other skills and ingest broad untrusted external content. The main risk is transitive trust: dev-scan and agent-council are not verified here, so this skill can inherit whatever permissions and data flows those downstream skills use. Optional Context7 appears same-org official and proportionate, but still adds normal supply-chain risk if installed via unpinned npx. No direct malware behavior or explicit credential theft is shown in this skill itself.