The Agent Skills Directory

[Indirect Prompt Injection Surface]: The skill is designed to ingest and process data from external files (e.g., CSV files like authors.csv). This creates a surface for indirect prompt injection where maliciously crafted data within these files could attempt to influence the agent's behavior during processing.
Ingestion Points: Data is read from local CSV files into staging tables or processed via Python/Node.js scripts.
Boundary Markers: The skill does not define specific delimiters for the CSV data content itself but strongly mandates the use of parameterized queries for all database operations.
Capability Inventory: The skill utilizes docker exec, sqlcmd, bcp, and database drivers (Node.js/Python) which can execute arbitrary SQL and interact with the file system.
Sanitization: The instructions explicitly require the use of parameterized statements and provide examples using ? placeholders or driver-specific parameter binding to prevent SQL injection.
[External Package Dependencies]: The skill instructs the user to install third-party packages from public registries, including @faker-js/faker, mssql (NPM), pyodbc, and faker (PyPI). These are well-known libraries used for generating test data and managing SQL connections.
[Command Execution]: The skill frequently uses docker exec to run administrative tools such as sqlcmd and bcp within the container. While necessary for the skill's purpose, this involves high-privilege operations within the container environment.
[Private Container Registry]: The skill references a private Azure Container Registry (sqldbpreview-dpgaeqhmgphzd4bk.azurecr.io) for the database image. It directs users to an official Microsoft signup page for credentials, which is consistent with its stated purpose of supporting a private preview product.

azuresql-db-seed