agent-framework-azure-ai-py

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill includes a runtime HTTP tool pointing at "https://learn.microsoft.com/api/mcp" (used in MCPStreamableHTTPTool) and an Azure AI service endpoint template "https://.services.ai.azure.com/api/projects/" (used via AZURE_AI_PROJECT_ENDPOINT) which are invoked at runtime and can directly influence agent inputs/outputs.