notion-plan

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

• Third-party content exposure detected (high risk: 1.00). This skill directly navigates to and scrapes user-supplied Notion pages via Playwright (Step 2: mcp__playwright__browser_navigate, browser_snapshot/browser_evaluate) and then parses and injects that extracted Notion content into the conversation and the downstream /design skill (Steps 3–5), so untrusted third-party page content can materially influence behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill navigates at runtime to a user-supplied Notion URL (e.g., https://www.notion.so/workspace/Page-Title-abc123), uses Playwright to fetch and extract the page content, and injects that content into the agent context and downstream /design call, so external Notion pages can directly control prompts.