wxo-builder

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs agents to ingest and use external, untrusted knowledge bases and APIs (see "Knowledge Base Providers" with external Milvus/AstraDB/Elasticsearch endpoints) and the custom Python tool example search_custom_vector_db which posts to https://my-custom-db.com/search, and those results are used by agents for RAG and answering user queries, so third-party content can materially influence agent behavior.