The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute a variety of dbt CLI commands such as dbt run, dbt build, dbt debug, and dbt docs serve. These commands perform system-level operations and can initiate local network services.
[EXTERNAL_DOWNLOADS]: The skill includes the dbt deps command, which is used to download and install external dbt packages from remote repositories like GitHub as defined in the project configuration.
[DATA_EXFILTRATION]: In the troubleshooting section, the skill suggests executing echo $DBT_ENV_SECRET_SNOWFLAKE_PAT. This command can result in the exposure of sensitive database credentials by printing them directly to the agent's output console or logs.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface through the processing of external data generated during dbt execution.
Ingestion points: The agent is instructed to read content from logs/dbt.log and compiled SQL files located in target/compiled/.
Boundary markers: No specific delimiters or instructions are provided to the agent to treat the content of these files as untrusted data or to ignore instructions embedded within them.
Capability inventory: The skill possesses the capability to execute shell commands and modify the file system through the dbt CLI.
Sanitization: There is no mechanism described for sanitizing or validating the contents of logs or compiled SQL artifacts before they are processed by the agent.

dbt-commands