spice-ai

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

• Third-party content exposure detected (high risk: 0.80). The skill exposes models to open web content via the configured websearch/web tool (see the "Web Search" example: tools:
• name: web from: websearch with engine: perplexity), which lets the agent fetch and read public third-party pages and thus could enable indirect prompt injection.