ai-code-walkthrough

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow explicitly ingests outsider-authored free text/code provided by the user at runtime (e.g., a pasted code block, a file path read on the user’s behalf, or a git diff/PR diff specified by the user), then the LLM explains it line-by-line—so an outsider can poison the walkthrough input directly.