refiner-automation

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly requires calling the external Rube MCP endpoint (https://rube.app/mcp) via RUBE_SEARCH_TOOLS to fetch tool schemas and recommended execution plans, and those third-party responses are read and used to determine tool selection and execution, so untrusted remote content can influence agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill requires connecting to the external MCP endpoint https://rube.app/mcp at runtime (via RUBE_SEARCH_TOOLS / RUBE_MANAGE_CONNECTIONS) to fetch tool schemas and recommended execution plans which directly influence agent prompts and execution flow, making this a required external dependency that can control the agent.