checkit

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md describes a runtime “Deep research” step that fetches external URLs/artifacts (outsider-authored web/repo/paper content) and then “analyze the sources”/“document the finding,” meaning the fetched free-form text is ingested into the agent’s LLM context for summarization and integration.