moai-mermaid-diagram-expert
Pass
Audited by Gen Agent Trust Hub on Mar 2, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: Indirect Prompt Injection surface via unsanitized Mermaid diagram input. \n
- Ingestion points: The
mermaid-to-svg-png.pyscript reads diagram source from external.mmdfiles. \n - Boundary markers: The script lacks delimiters or safety instructions when processing external diagram code. \n
- Capability inventory: The conversion script uses Playwright (headless browser) which executes JavaScript and writes to the local filesystem. \n
- Sanitization: The Mermaid renderer is configured with
securityLevel: 'loose', which allows script execution, and diagram content is injected into HTML f-strings without escaping. \n- [COMMAND_EXECUTION]: The skill uses subprocess calls to execute CLI tools. \n - Evidence:
subprocess.run(['mmdc', ...])in theSKILL.mdexample. \n- [EXTERNAL_DOWNLOADS]: Fetches external JS libraries during rendering. \n - Evidence: Downloads the Mermaid.js library from the jsDelivr CDN, a well-known service.
Audit Metadata