churn-winback-leadgen
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references the ManyPI platform and provides a URL to its MCP server endpoint (https://mcp.manypi.com/mcp) for tool integration.- [COMMAND_EXECUTION]: Includes a shell command to add a Model Context Protocol (MCP) server to the agent environment.- [INDIRECT_PROMPT_INJECTION]: The workflow relies on ingesting external data (research signals about customer accounts) to drive the creation of win-back messages, which represents an indirect prompt injection attack surface.
Audit Metadata