ai-deep-research

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

• Third-party content exposure detected (medium risk: 0.30). SKILL’s runtime workflow is explicitly built for web/public-source deep research: the searcher executes queries and reads/browses retrieved source pages, extracting free text into the SourceLedger for later synthesis, which means outsider-authored content can be ingested from sources selected via the plan/search loop.