audit

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

• Third-party content exposure detected (high risk: 0.75). Outsider free text can enter the LLM context via the local LLM (llm) detection layer inside core.anonymize(text, cfg) / core.anonymize(res["redacted_text"], cfg), where text is read from user-supplied files (e.g., arbitrary .md/.txt paths from --list or a folder) and then passed to the LLM for detection; if those files contain outsider-authored content, it becomes LLM-readable input.