mcp-builder
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The
scripts/evaluation.pyandscripts/connections.pyfiles provide functionality to execute local shell commands via thestdiotransport. This is an intended capability designed to allow developers to launch and test their own MCP server implementations locally. - [PROMPT_INJECTION]: The evaluation script processes input from XML files and passes it to the AI model. This presents a surface for indirect prompt injection if a developer runs an evaluation file from an untrusted source. The script includes structural mitigations by wrapping model outputs in XML tags for parsing.
- [EXTERNAL_DOWNLOADS]: The skill instructions direct the agent to fetch documentation from official sources, including the
modelcontextprotocol.iowebsite and the Model Context Protocol GitHub repositories. These are well-known and trusted sources for the protocol's technical documentation.
Audit Metadata