llm-wiki-eval

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md describes evaluating “with-wiki” using retrieved wiki pages and raw sources (including untrusted context like PDFs/emails/tickets/chat transcripts) to compute security/prompt-injection outcomes, meaning outsider-authored free text can be ingested via the retrieval/index pipeline when the workflow evaluates recent or selected sources.