extract-source-sample

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow reads and dumps verbatim the contents of HOW_TO_MAKE_THIS_VIDEO.md from the provided run folder into the agent’s LLM context (free-form markdown authored by someone other than the operating user), creating an indirect prompt-injection path.