clause-analyze

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s workflow explicitly allows “Read relevant existing documents from the project” and “Use WebSearch/WebFetch for current regulatory guidance,” which at runtime can ingest outsider-authored free text from public web pages or externally provided documents into the agent’s LLM context via those fetch/read paths.