The Agent Skills Directory

Indirect Prompt Injection Surface: The skill provides tools and instructions for reading from and writing to Azure Storage services (Blobs, Queues, Tables, and Data Lake). This represents a standard attack surface where the agent might ingest untrusted data that could contain instructions intended to influence its behavior.
Ingestion points: Data is brought into the agent's context through tools like storage_blob_get and queue message retrieval described in SKILL.md and SDK references.
Capability inventory: The skill possesses the ability to list resources, download data, and upload content across various Azure Storage services.
Boundary markers: The skill includes a dedicated auth-best-practices.md file that strongly emphasizes the use of Azure RBAC and Managed Identities to enforce the principle of least privilege, which serves as a significant structural mitigation.
Sanitization: The instructions focus on API usage; developers implementing these patterns should ensure that data retrieved from storage is sanitized before being used in downstream prompts.

azure-storage