creating-diagrams

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill's draw.io MCP tools explicitly accept a content parameter that can be "ダイアグラム内容、またはコンテンツへのURL" (references/DRAWIO-FORMATS.md), so the agent may fetch and ingest arbitrary public URLs (Mermaid/XML/CSV) containing untrusted, user-generated content (including CSV directives or Mermaid click/links) which the agent is expected to read/interpret when building/rendering diagrams.