llm-wiki

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

• Third-party content exposure detected (medium risk: 0.30). The required ingest runtime flow reads the user-provided source file content from the vault’s raw/ directory (e.g., via /wiki-ingest <path> → references/ingest-workflow.md step 2 → LLM read tool), so outsider-authored free text can enter the model at ingestion time.