assumption-surface

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

• Third-party content exposure detected (high risk: 0.70). The skill’s runtime workflow ingests the user-provided “artifact” text (the memo/claim being stress-tested) into the LLM context; if that artifact contains outsider-authored free text (e.g., a web page, issue/PR body, or other third-party message), that text becomes LLM-readable prose and is fed into the model.