cawcut-generate

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow runs cawcut app list --json and uses the returned App name/description to decide whether to hand off to cawcut-app-run, so BE-fetched catalog text (outsider-authored by third-party App publishers) can enter the agent’s LLM context via the tool outputs and subsequent prompt construction.