The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The skill references specific CLI commands like af dags list and af dags source. These are consistent with the intended purpose of managing Airflow-like DAG environments and do not perform unauthorized or dangerous operations.
[DATA_EXPOSURE & EXFILTRATION] (SAFE): The skill provides SQL queries to inspect metadata (e.g., information_schema.view_table_usage). This is appropriate for a lineage tracing tool and does not attempt to exfiltrate data to external domains.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill ingests external data in the form of DAG source code and database metadata. While this represents a theoretical attack surface where malicious instructions could be embedded in code comments, the skill is instructional and does not demonstrate exploitable capabilities for this vector.

tracing-downstream-lineage