home-depot-search

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

• Third-party content exposure detected (high risk: 0.75). The required workflow calls curl to fetch a runtime HTTP response from https://ai-factory.frevana.com/..., then reads that response body as text (python3 ... read_text(...)) and finally prints it to stdout (cat "$RESULT_FILE"), so outsider-authored free text from the remote API can enter the agent/LLM context.