skill

SUSPICIOUS: the core purpose is coherent, and the named official MCP examples largely match legitimate docs, but the skill’s footprint is broader than a simple setup helper because it persistently installs external MCP servers, forwards credentials to them, and offers a custom mode for arbitrary commands or remote URLs. This is not confirmed malware, but it carries medium supply-chain and credential-forwarding risk.

Mostly coherent CI/CD helper with low supply-chain concern and no obvious credential theft path. The main risk is FORCE MODE, which lets the agent auto-commit and push code to trigger external pipelines without confirmation, making this a medium-risk automation skill rather than malware.