mcp-builder
Pass
Audited by Gen Agent Trust Hub on Feb 21, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS] (LOW): The skill instructions in SKILL.md guide the agent to fetch documentation from modelcontextprotocol.io and GitHub. While these are trusted sources for the protocol, fetching external content to guide logic introduces a dependency on remote integrity.
- [COMMAND_EXECUTION] (LOW): The utility script connections.py implements a stdio transport that executes local commands, which is inherent to the MCP specification. The guide also instructs the agent to run testing tools like the MCP Inspector and python compiler.
- [PROMPT_INJECTION] (LOW): The skill exhibits an indirect prompt injection surface by ingesting external markdown data from URLs to guide code implementation. Evidence: Ingestion points in SKILL.md (READMEs and sitemaps); Boundary markers are absent; Capability inventory includes command execution and file writing; Sanitization is not documented for the fetched content.
Audit Metadata