helix-mcp-craft-content
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill connects to a hosted MCP endpoint at 'https://craft-content.gtmhelix.com/mcp' to provide its primary functionality. This domain is associated with the skill's author ('shashwatgtm').
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted user-supplied data, such as customer win metrics, drafts, and testimonials, to generate content. This creates a surface for indirect prompt injection where instructions could be hidden in processed content. However, the skill includes rigorous 'Operating Principles' (in references/operating-principles.md) that mandate fact-checking, assumption flagging, and verification of claims using 'web_search' and 'web_fetch' tools, which serves as a mitigation strategy.
- [COMMAND_EXECUTION]: The documentation provides CLI instructions for registering the MCP server using 'claude mcp add'. These are standard configuration commands for the platform and do not represent a security risk.
Audit Metadata