The Agent Skills Directory

[SAFE]: The skill consists of documentation and configuration templates for dbt (data build tool), focusing on project structure and data integrity.
[EXTERNAL_DOWNLOADS]: The CI configuration example in .github/workflows/dbt-ci.yml references official GitHub Actions (actions/checkout@v4, actions/setup-python@v5) and the official dbt-snowflake package from the Python Package Index. These are well-known and trusted resources in the data engineering ecosystem.
[COMMAND_EXECUTION]: The skill documents standard dbt CLI operations (dbt deps, dbt build, dbt test, dbt snapshot). These commands are essential for the primary purpose of managing a data warehouse project and are used within documented examples.
[INDIRECT_PROMPT_INJECTION]: As a data transformation tool, dbt projects naturally ingest data from external sources (e.g., Stripe API data in the examples). While this represents a standard vulnerability surface for data pipelines (Cat 8), the skill follows best practices by using staging layers and specific source definitions to control data flow.
Ingestion points: Database source tables defined in __sources.yml (e.g., raw_stripe).
Boundary markers: dbt's source() and ref() functions act as logical boundaries between raw data and modeled logic.
Capability inventory: The skill uses SQL and Jinja templates to execute transformations on a target data warehouse.
Sanitization: Standard SQL parameterization and database driver escaping are managed by the underlying dbt-core engine.

dbt-data-modeling