mushi-integration

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

• Third-party content exposure detected (medium risk: 0.30). Stages 3–5 ingest outsider-authored free text from a live URL (e.g., mushi stories map --url https://your-app.com --wait crawls pages) that the LLM uses for story mapping and then uses those generated artifacts for TDD/test generation and execution.

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

• Potentially malicious external URL detected (high risk: 0.80). The command "mushi stories map --url https://your-app.com --wait" (https://your-app.com) fetches live site content at runtime which is then used as input for Claude to draft user stories, so the external URL can directly control prompts.