The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on a bash tool to execute the altimate-dbt CLI. This tool performs critical project operations including compiling code, building models, and running tests. It also includes an execute command that runs arbitrary SQL queries against the connected data warehouse.
[EXTERNAL_DOWNLOADS]: The skill uses altimate-dbt deps and altimate-dbt add-packages to manage project dependencies. These commands download and install external dbt packages from configured registries, which is a standard part of dbt project initialization and maintenance.
[INDIRECT_PROMPT_INJECTION]: The skill processes external SQL model files and YAML configuration files, creating a surface where malicious instructions could theoretically be embedded in data processed by the agent.
Ingestion points: SQL models (models/**/*.sql) and YAML metadata (schema.yml, sources.yml) are read into context during development and validation steps (SKILL.md, references/yaml-generation.md).
Boundary markers: None detected; the agent is instructed to read these files directly to understand project logic and conventions.
Capability inventory: The skill possesses extensive capabilities including file-write/edit access, bash execution for the altimate-dbt toolset, and the ability to execute arbitrary SQL on the warehouse via altimate-dbt execute (references/altimate-dbt-commands.md).
Sanitization: No explicit sanitization or instruction-filtering of the ingested SQL or YAML content is described before the agent processes it.

dbt-develop