exploring-the-wizard

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

• Third-party content exposure detected (high risk: 0.70). The required workflow uses wizard-ci MCP tools like read_state/render_screen to read and render the live wizard’s screens from the user-provided app runtime (outsider-authored free text from the app’s UI/content), which is then ingested into the agent’s LLM context.