diagram

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

• Third-party content exposure detected (medium risk: 0.30). Source: the skill’s workflow step “Extract nodes and edges — read labels… from the user’s request or conversation context”, which implies outsider-authored free text is ingested to create node/edge labels that are then rendered into /tmp/diagram.html and awaited for screenshot.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The template includes a runtime tag that fetches and executes remote JavaScript from https://cdnjs.cloudflare.com/ajax/libs/cytoscape/3.29.2/cytoscape.min.js, which the skill requires to render diagrams.