drawio-diagram

Warn

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses npx to dynamically download and execute the @drawio/mcp package from the npm registry. While draw.io is a well-known service, executing remote packages at runtime introduces a dependency on external infrastructure.
  • [REMOTE_CODE_EXECUTION]: The use of 'npx -y @drawio/mcp' within the MCP server call constitutes the execution of remote code on the local system.
  • [COMMAND_EXECUTION]: The skill provides instructions to construct and run shell commands for exporting diagrams (using the draw.io CLI) and opening files (using open, xdg-open, or start). These commands use filenames and export formats derived from user requests, which could lead to command injection if the agent does not strictly validate the input characters.
  • [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it processes untrusted diagram content (Mermaid, XML, and CSV) which could contain embedded instructions.
  • Ingestion points: User-provided content in the 'open_drawio_mermaid', 'open_drawio_xml', and 'open_drawio_csv' tool calls.
  • Boundary markers: No explicit delimiters or 'ignore instructions' warnings are defined for the diagram data.
  • Capability inventory: Includes local file system write access, shell command execution for file management and export, and network-based package execution via npx.
  • Sanitization: The skill does not specify any sanitization or escaping procedures for the content before it is processed by the CLI or browser-based tools.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 6, 2026, 12:31 AM