mcp-specialist

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to discover and provision MCP servers from the public ecosystem (e.g., "search MCP ecosystem" and NPX commands in prompts/provision-server.prompt.md and SKILL.md, plus docker image pulls), run them, and read their initialize/tools responses—untrusted third‑party package/page content that the agent ingests and uses to decide tool mappings and actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly runs remote packages/images at runtime (e.g., "npx -y @microsoft/playwright-mcp" and "docker pull mcr.microsoft.com/playwright/mcp:latest"), which fetches and executes remote code and are presented as required deployment options for the MCP servers.