codex-image-gen

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

• Third-party content exposure detected (medium risk: 0.30). The runtime path scripts/codex_image.py ingests outsider-provided free text prompts via --prompt / --prompt-file and embeds them into the request body (build_request_body → input_text), which is then sent to the Codex Responses API, so an outsider can directly supply the free text that the workflow uses without selecting a trusted item first.