retirement-income

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime path calls the public planfi MCP tools (e.g., analyze_withdrawal_strategy, optimize_social_security, etc.) over https://ai.planfi.app/mcp/free, and the LLM ingests the tools’ returned structured outputs (including assumed_defaults[] and next_actions[]) as readable text—these tool responses are outsider-authored content from a third-party service.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill is a thin orchestration layer that at runtime calls the PlanFi MCP endpoints (https://ai.planfi.app/mcp/free and https://ai.planfi.app/mcp), and the MCP's returned outputs (assumed_defaults[] and next_actions[]) directly drive agent instructions and tool-chaining, so the external URL controls prompts/behavior.