The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and generate dbt project components, which theoretically exposes an attack surface for instructions embedded in data models or configuration files.
Ingestion points: The skill instructions involve reading and defining model layers, staging definitions, and schemas from a dbt project context.
Boundary markers: There are no explicit delimiters provided in the patterns to differentiate between transformation logic and potential malicious instructions in source data metadata.
Capability inventory: The agent is authorized to generate SQL transformations and dbt configurations based on these patterns.
Sanitization: Not explicitly implemented within the patterns, as it relies on the dbt framework's native handling of SQL and YAML.
[COMMAND_EXECUTION]: The implementation playbook documents standard dbt CLI commands used for development and production runs.
Evidence: Commands such as dbt run, dbt test, dbt build, and dbt docs generate are listed in resources/implementation-playbook.md. These are routine operations for the tool's intended use case.
[SAFE]: The skill implements best practices for analytics engineering, including the medallion architecture (staging, intermediate, and marts layers). It references the widely-used and trusted dbt-utils package and official dbt documentation. No obfuscation, data exfiltration patterns, or hardcoded credentials were identified.

dbt-transformation-patterns