visualize

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

• Third-party content exposure detected (high risk: 0.85). The required runtime path reads outsider-authored Markdown files from the target OKF bundle directory (bundle.rglob("*.md") → p.read_text(...) → embedded into the generated HTML as n.body and rendered via marked.parse(n.body)), so any non-user-authored .md content becomes free text in the agent’s LLM context (indirectly via the browser-rendered HTML content).

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

• Potentially malicious external URL detected (high risk: 0.70). The generated HTML embeds external scripts that are fetched and executed when the output is opened in a browser (required for the visualization): https://cdn.jsdelivr.net/npm/cytoscape@3.30.2/dist/cytoscape.min.js and https://cdn.jsdelivr.net/npm/marked@14/marked.min.js.