mermaid-mcp

Fail

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The installation process requires the global installation of the @narasimhaponnada/mermaid-mcp-server package from the npm registry. This package originates from an unverified third-party developer.
  • [REMOTE_CODE_EXECUTION]: The skill configuration requires the agent to execute a JavaScript file (index.js) located within the downloaded third-party package using the Node.js runtime. This execution of unverified external code represents a significant security risk.
  • [COMMAND_EXECUTION]: The setup instructions involve the execution of several shell commands, including npm install, npm root, and the spawning of node processes to run the MCP server.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted diagram code (Mermaid syntax) through its tools.
  • Ingestion points: Data enters through the generateDiagram and validateDiagram tools defined in SKILL.md.
  • Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the tool definitions.
  • Capability inventory: The skill possesses file-writing capabilities (rendering SVGs) and syntax validation as described in the tools table.
  • Sanitization: There is no evidence of input sanitization or validation of the Mermaid syntax before it is processed by the external server.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 11, 2026, 12:32 PM