lead-sourcing
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the live web, creating a potential attack surface for indirect prompt injection.
- Ingestion points: The agent is instructed to search the web and visit candidate sites to extract lead data (SKILL.md).
- Boundary markers: The instructions lack specific boundary markers or instructions to disregard potential malicious commands embedded in external web data.
- Capability inventory: The skill uses a vendor-provided MCP server (https://mcp.manypi.com/mcp) capable of web searching, site scraping, and email outreach.
- Sanitization: No sanitization or validation mechanisms are described for the content retrieved from external sites.
- [EXTERNAL_DOWNLOADS]: The skill instructions include a command to add a remote Model Context Protocol (MCP) server:
claude mcp add --transport http manypi https://mcp.manypi.com/mcp -s user. This connects the agent to the vendor's external API.
Audit Metadata