building-ai-agent-on-cloudflare

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

• Third-party content exposure detected (high risk: 0.90). The skill's Agent Patterns show agents fetching and ingesting data from external public sources (e.g., the ToolAgent handler fetching https://api.weather.com/${city}, the RAGAgent ingestDocument/Vectorize flows that build context from document.metadata.source, and MCP client connections to external SSE endpoints in references/agent-patterns.md), and those third-party results are fed back into the AI and used to decide tool calls and responses, which clearly exposes the agent to untrusted user-generated or public web content that can influence behavior.