knowledge-index

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

• Third-party content exposure detected (high risk: 0.70). This workflow reads outsider-authored free text from the knowledge base on disk—e.g., it renders sub-indexes/root via sub_index.py/root_index.py using <wiki_path> and the wiki’s existing page content (which can include external user-authored edits), and that content is ingested into the LLM context during those render steps.