skills/microsoft/skills/agents-v2-py/Gen Agent Trust Hub

agents-v2-py

Pass

Audited by Gen Agent Trust Hub on Apr 28, 2026

Risk Level: SAFE
Full Analysis
  • Standard Package Installation: The instructions include the installation of azure-ai-projects and azure-identity. These are official Microsoft SDKs required for the functionality described and originate from a trusted vendor environment.
  • Secure Authentication Patterns: The documentation consistently recommends using DefaultAzureCredential, which is a security best practice for Azure development. This approach avoids the need for hardcoded credentials by leveraging Managed Identities or local development environments (like Azure CLI or Environment Variables).
  • External Resource Configuration: The skill allows for the configuration of external Model Context Protocol (MCP) servers and container registries. While these represent network touchpoints, they are documented as user-defined parameters for integrating the user's own infrastructure and services.
  • Indirect Prompt Injection Surface: By facilitating the creation of agents that ingest data from external tools (such as MCP servers or custom containerized logic), the skill manages an inherent surface for indirect prompt injection.
  • Ingestion points: Data enters the agent context via MCP tool outputs (server_url) and container responses.
  • Boundary markers: Not explicitly defined in the basic code templates provided.
  • Capability inventory: The agent is configured with code_interpreter, file_search, and custom MCP tools as described in SKILL.md.
  • Sanitization: Standard sanitization is not demonstrated in these configuration examples; users should implement validation within their custom agent logic as a secondary security measure.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 28, 2026, 03:16 PM