The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from multiple external sources, which creates a surface for potential indirect prompt injection attacks.
Ingestion points: The agent is instructed to ingest and process results from dbt show (SKILL.md, references/discovering-data.md), general warehouse queries, and JSON metadata from the hub.getdbt.com API (references/managing-packages.md).
Boundary markers: The skill contains explicit instructions in SKILL.md ("Handling external data") and references/managing-packages.md for the agent to ignore instruction-like text embedded in data and to treat all external inputs as untrusted.
Capability inventory: The agent has access to Bash, Write, and Edit tools, which could be exploited if the agent were to obey instructions found in external data.
Sanitization: The skill relies on natural language instructions for the agent to extract only structured fields and ignore executable content, but no programmatic sanitization or strict schema validation is enforced by the skill code itself.
[EXTERNAL_DOWNLOADS]: The skill references the official dbt package registry for discovery and installation.
Evidence: references/managing-packages.md provides URLs for the dbt Hub API at https://hub.getdbt.com/api/v1/index.json. This is the official registry maintained by the skill author (dbt Labs) and is used for standard package management via dbt deps.

using-dbt-for-analytics-engineering