material-ui-docs

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). At runtime the skill’s workflow consults the bundled Material UI reference markdown files (e.g., reference/*.md), which are derived from public web pages on mui.com and thus constitute outsider-authored free text being ingested into the agent context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The MCP setup instructs running remote code at runtime via "npx -y @mui/mcp@latest" (and related npx inspector commands) which would fetch/execute a remote package and the MCP rules explicitly direct the agent to fetch additional docs/URLs at runtime, meaning external content can control agent prompts and behavior.