viking-knowledgebase

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

• Third-party content exposure detected (high risk: 0.80). The skill's scripts/search.py sends the user's query to the external Viking knowledge-base API at api-knowledgebase.mlp.cn-beijing.volces.com and prints/returns result.get("content"), meaning the agent will fetch and ingest third-party knowledge-base content (potentially user-generated) as part of its workflow.