The Agent Skills Directory

[Secure Authentication Patterns]: The skill strongly emphasizes the use of DefaultAzureCredential, which avoids hardcoding sensitive API keys or connection strings. It explicitly advises against using static credentials that bypass audit and rotation protocols.
[Credential Scoping]: It provides guidance on using the require_envvar parameter and specific credential classes like ManagedIdentityCredential to constrain the credential chain in production environments, reducing the risk of accidental credential leakage from local developer environments.
[Resource Lifecycle Management]: The instructions mandate the use of Python context managers (with and async with) for both clients and credentials. This ensures that network transports, sockets, and token caches are deterministically released, preventing potential resource exhaustion or data leakage in memory.
[Safe Data Handling]: While the skill includes examples of downloading artifacts (e.g., download_blob), it uses standard file operations with locally defined paths. There is no evidence of path traversal vulnerabilities or execution of downloaded content.
[Environment Variable Usage]: The skill uses environment variables for configuration (AZURE_CONTAINERREGISTRY_ENDPOINT), which is a standard and safe practice for cloud-native application configuration.

azure-containerregistry-py