The Agent Skills Directory

[CREDENTIALS_UNSAFE] (SAFE): The skill identifies and warns against hardcoding credentials. Code examples use clearly marked placeholders (e.g., <tenant-id>, <client-secret>) or well-known public constants for local emulators (e.g., Azurite/MinIO defaults), which pose no risk to production environments.
[DATA_EXPOSURE_AND_EXFILTRATION] (SAFE): The documentation explicitly instructs users on how to protect sensitive files like .env and service account keys. It promotes the use of detect-secrets and .gitignore to prevent accidental exposure.
[EXTERNAL_DOWNLOADS] (SAFE): All mentioned dependencies are standard, high-reputation libraries (e.g., boto3, azure-identity, google-cloud-storage) and tools from trusted sources. No suspicious remote code execution or piped-shell installations are present.
[COMMAND_EXECUTION] (SAFE): The provided CLI commands (AWS CLI, GCP GCloud, Azure CLI) are standard administrative tasks for configuring identity and access management and do not contain malicious payloads.
[PROMPT_INJECTION] (SAFE): No instructions attempting to override agent behavior or bypass safety filters were detected in the markdown or code comments.

data-engineering-storage-authentication