The Agent Skills Directory

Indirect Prompt Injection Surface: The skill references tools and SDK methods that ingest external data (documents, images, audio URLs, and search results) for AI processing. This represents a potential surface where content embedded in that data could attempt to influence agent behavior. Evidence includes methods like analyzeFromUrl and content_urls used for analyzing remote media across the provided reference guides.
External Dependency References: The skill documents the installation of official Azure SDK packages via standard package managers like npm and pip. These are official resources from a well-known vendor and are used within the scope of their intended functionality. Evidence includes instructions for @azure/identity, azure-search-documents, and azure-ai-contentsafety.
Command Line Interaction: The skill mentions usage of the Azure CLI (az) for specific management operations, which is a standard practice for cloud environment configuration. Evidence includes mentions of az search and az cognitiveservices in SKILL.md.
Authentication Best Practices: The skill includes proactive security guidance, specifically discouraging the use of static credentials and promoting Azure Managed Identities and Role-Based Access Control (RBAC). Evidence is found in the dedicated references/auth-best-practices.md guide.

azure-ai