excalidraw-skill

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructions provide guidance on cloning a repository from the author's GitHub account (https://github.com/yctimlin/mcp_excalidraw) to set up the necessary backend server. As this is a vendor-owned resource, it is a standard deployment procedure for the skill.- [COMMAND_EXECUTION]: The skill includes several Node.js utility scripts (e.g., scripts/clear-canvas.cjs, scripts/create-element.cjs) that are intended to be executed locally. these scripts perform standard operations such as making HTTP requests to a local API and writing exported elements to a specified file path.- [PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection through its ingestion of diagram data and Mermaid syntax.
  • Ingestion points: Elements are read from the server via the describe_scene tool and from local files via import-elements.cjs.
  • Boundary markers: No specific boundary markers or safety instructions are present to encapsulate external data.
  • Capability inventory: The skill can perform network requests and local file system writes.
  • Sanitization: No explicit sanitization of canvas elements or Mermaid syntax is performed before the data is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 05:24 AM