apache-airflow-orchestration

Pass

Audited by Gen Agent Trust Hub on May 16, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructions fetch Airflow installation constraints from the official Apache Airflow GitHub repository and download a Docker Compose configuration from the official Airflow documentation domain.- [COMMAND_EXECUTION]: Provides implementation patterns for BashOperator and DockerOperator, which allow for shell command execution and container management as part of standard DAG orchestration.- [PROMPT_INJECTION]: The skill demonstrates how to ingest data from external untrusted sources such as web APIs, S3 buckets, and local files, and then process this data using tasks with code execution capabilities. This represents an indirect prompt injection surface.
  • Ingestion points: External data is accessed via requests.get in custom sensors, FileSensor for local files, and S3Hook for cloud storage.
  • Boundary markers: The provided code examples do not include explicit delimiters or instructions to ignore embedded commands within the processed data.
  • Capability inventory: The skill documentation covers the use of BashOperator, PythonOperator, and DockerOperator, which provide significant execution privileges within the environment.
  • Sanitization: There are no examples of input validation, escaping, or sanitization for the content retrieved from external sources before it is used in subsequent tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
May 16, 2026, 02:37 PM
Security Audit — agent-trust-hub — apache-airflow-orchestration