The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it is designed to ingest and process user-provided dbt project artifacts to generate code and configurations.
Ingestion points: Processes user-provided SQL model files and YAML configuration files (e.g., _sources.yml, _models.yml) as mentioned in SKILL.md and multiple reference files.
Boundary markers: Absent. The skill's instructions do not define specific delimiters or ignore-instructions markers for data read from external project files.
Capability inventory: The skill can generate complex SQL logic, dbt CLI commands, and automated CI/CD workflow configurations across all reference guides.
Sanitization: Absent. There are no instructions for sanitizing or escaping user-provided project content before it is used by the agent to generate new code.
[EXTERNAL_DOWNLOADS]: Recommends the installation of dbt adapter packages and utility packages from standard registries.
Evidence: The skill suggests installing dbt-snowflake, dbt-bigquery, and dbt-duckdb via pip in ci-cd-deployment.md, consulting-workflow.md, and duckdb-adapter.md. It also references the use of well-known community packages like dbt-utils and dbt-expectations in jinja-macros-packages.md.
[COMMAND_EXECUTION]: Instructs the agent on the use of standard dbt CLI commands to manage the data transformation lifecycle.
Evidence: References to dbt build, dbt run, dbt test, and dbt compile are found throughout SKILL.md and the reference documents as standard operational procedures for dbt projects.

dbt-transforms