prompt-engineer-toolkit

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/prompt_tester.py, the required workflow can read outsider-authored free text at runtime via JSON/stdin --input (cases) and case["input"], which are then inserted into prompts (and optionally into an external --runner-cmd) without first selecting a specific trusted item.