cadence-check

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow reads outsider-authored free text from files under context/ at runtime (e.g., index.json via json.loads(...) and *.md via path.read_text(...) in _read_text_safe and _gather_learnings_text), then includes derived strings (like parse failures/warnings containing file contents paths and parse issues) into the returned summary.scan_metadata, which is later formatted and persisted by downstream user-facing wrappers.