moai-cc-mcp-builder

The package is a plausible legitimate AI-assisted MCP development skill with heavy Context7 integration. I found no explicit malicious code, hard-coded credentials, or direct exfiltration payloads in the supplied content. However, the declared operational capabilities (Read, Write, Bash, WebFetch) combined with CI/CD integration examples create a moderate supply-chain and privilege-abuse risk: if the skill (or transitive moai-* dependencies) is executed inside developer machines or CI runners that hold secrets, generated code or automation steps could accidentally or intentionally expose sensitive data or execute unwanted commands. Recommended mitigations: run generation and tests in isolated environments, require explicit interactive approvals before writing files or executing shell commands, restrict network and file access (least privilege), audit Context7 endpoints and all transitive moai-* skills before adding them to CI, and add explicit sanitization/review gates for any externally fetched patterns used to generate executable artifacts.