wiki-ingest

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

• Third-party content exposure detected (high risk: 0.85). The skill’s workflow ingests external files via the Read tool in Step 1 (“For external files, use the Read tool”), and then reads/extracts their body text in Step 3 (“Read the source content”), which becomes LLM-readable free text in the agent context—i.e., outsider-authored document content.