simulation

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

• Third-party content exposure detected (medium risk: 0.30). Bounded-context discovery (--discover-bcs) reads ALL board item text from a live Miro board via miro_list_board_items (or parses a directly supplied board export), so an outsider can supply/poison the free text that the required workflow ingests without selecting a specific item first.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill accepts and (when Miro is available) reads a user-supplied Miro board URL at runtime (e.g., "--discover-bcs https://miro.com/app/board/abc123"), fetching board contents via MCP and using that data to drive agent prompts/analysis, so the external URL is used during runtime to control agent behavior.