learn

SUSPICIOUS: the core learnings functions are plausible, but the skill is internally inconsistent because it embeds unrelated routing/migration automation and auto-commit behavior far beyond its stated purpose. The gstack binaries appear same-org and publicly documented, so this is not confirmed malware, but the scope mismatch and eval-driven shell execution make it a medium-to-high security concern for an AI skill.